# DidReset — full reference (llms-full.txt) > DidReset is an independent tracker of official, account-wide usage-limit resets for OpenAI Codex, Anthropic Claude Code and xAI Grok. This file contains the answer capsule, definitions, tables, FAQ and sources of every reference page, in English and Chinese. Not affiliated with OpenAI, Anthropic or xAI. Cadence estimates are estimates, not vendor promises. Generated 2026-09-17 from src/content: 28 error pages, 24 guides, 3 limits pages. Site: https://didreset.com ## Live pages (data changes every 60 seconds; read the JSON API for current values) - Departures board: https://didreset.com/ · JSON: https://didreset.com/api/status · RSS: https://didreset.com/feed.xml - Codex reset history: https://didreset.com/codex · JSON: https://didreset.com/api/timeline?provider=codex - Claude reset history: https://didreset.com/claude · JSON: https://didreset.com/api/timeline?provider=claude - Grok reset history: https://didreset.com/grok · JSON: https://didreset.com/api/timeline?provider=grok ## Errors, by exact message ### "You've hit your usage limit. Upgrade your plan or add credits to continue, or try again at {Mon D, YYYY, H:MM AM}." — Codex URL: https://didreset.com/errors/codex-you-have-hit-your-usage-limit-upgrade-or-add-credits-try-again-at Language: en · Surface: App · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade your plan or add credits to continue, or try again at {Mon D, YYYY, H:MM AM}. "You've hit your usage limit. Upgrade your plan or add credits to continue, or try again at {Mon D, YYYY, H:MM AM}." is Codex's message when one of your two windows, the 5-hour window or the weekly limit, is used up. It refreshes on its own clock, shown by /status and Settings → Usage; it is not the account-wide reset this board tracks. The last official Codex reset on record was {lastResetDate} (source linked in the board row). What does "You've hit your usage limit. Upgrade your plan or add credits to continue, or try again at {Mon D, YYYY, H:MM AM}." mean? This is the current Codex limit notice for personal ChatGPT plans, seen in the desktop app and in the same wording family in the CLI and IDE extension. OpenAI meters Codex over two windows, "a five-hour window and a weekly window", and the time after "try again at" is when the exhausted window refreshes for your account. The three options in the sentence are the ones OpenAI's help center lists: "You may be able to add credits, use an available reset, upgrade, or wait until the displayed reset time." Eligible Plus and Pro users can buy credits without changing plans. The source report (openai/codex issue #34360, 2026-07-20) came from a Pro user on Codex app 26.715.52143 for Windows: the app showed 99% remaining, and redeeming a banked weekly reset only replaced the timestamp with a new one (Jul 27, 2026, 10:26 AM). OpenAI labelled it bug / rate-limits, so the message can appear while your own bars still show room. Which window is full? - Window: 5-hour window · Resets: Five hours after the first message that opened the window, per account. OpenAI: "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." Removed for Plus, Business and Pro on 2026-07-12; back for Plus since 2026-08-25 (the post says "Tomorrow" and is stamped 2026-08-25 01:16 UTC, i.e. Pacific time 2026-08-25); "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. · Where to check: /status inside an active Codex CLI session (the "5h limit" line); Settings → Usage in the app; chatgpt.com/codex/settings/usage - Window: Weekly limit · Resets: Once a week on your account's own date. OpenAI publishes no weekly figure ("Weekly limits may also apply"). A full banked reset "changes your weekly reset date"; after a purchased reset the next automatic weekly reset is seven days after your first request. · Where to check: /status (the "Weekly limit" line); Settings → Usage; chatgpt.com/codex/settings/usage What to do when you see it 1. Find out which window is full: In an active Codex CLI session type /status, or open Settings → Usage in the app. OpenAI's help center: "Check which allowance is exhausted, your credit balance, and any reset time shown." The "5h limit" and "Weekly limit" lines each carry their own reset time. 2. Check for a banked reset: If the profile menu shows "1 reset available", redeeming a full banked reset "refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date". It is consumed only when it actually refreshes at least one window. 3. Add credits, upgrade, or wait: Eligible Plus and Pro users can buy credits without changing plans; credits pay for additional usage and are not a reset. Otherwise the window refreshes at the time in the message. 4. Report it if the numbers disagree: If /status or Settings → Usage still shows allowance in both windows, file an issue at github.com/openai/codex with the /status output (label rate-limits), or contact OpenAI Support with the reset time shown, a screenshot, the client and model, and your time zone. When does it reset? Your own window: the time after "try again at" is your account's refresh time for the window that ran out. The 5-hour window rolls five hours after the message that opened it; the weekly limit refreshes on your account's own weekly date. Paste that time into the window tracker to get a countdown and a calendar file. Account-wide resets: OpenAI occasionally announces a reset for everyone; "An automatic reset is applied directly to the usage limits covered by the announcement." Those are the events on this board and on the Codex history page. They are not scheduled, so the board only shows the last one and a cadence estimate. FAQ: Q: Which window does "try again at" refer to? A: The one that ran out. Codex tracks a 5-hour window and a weekly limit per account; the message does not say which, but /status prints a separate reset time next to "5h limit" and "Weekly limit". A time hours away is usually the 5-hour window; a date days away is the weekly limit. Q: I redeemed a banked reset and still see this message. Why? A: That is the bug behind issue #34360: a Pro user at 99% remaining used a weekly reset and only got a new "try again at" timestamp. OpenAI says a banked reset "is consumed only when it successfully refreshes at least one eligible usage window", so check Settings → Usage for the updated reset time and report it if the reset was spent without refreshing anything. Q: Can OpenAI Support reset my usage limit? A: No. The help center is explicit: "OpenAI Support does not reset ChatGPT or Codex usage limits. If you reach a limit, wait until it resets or use another available option shown in your account." Support does investigate if your usage was counted incorrectly or access has not returned after the stated reset time. Q: Does an account-wide Codex reset clear this message? A: When OpenAI announces one, yes: an automatic reset "is applied directly to the usage limits covered by the announcement" and does not need to be redeemed. The board row above shows whether such a reset happened in the last 24 hours; it cannot predict the next one. Sources: 1. openai/codex issue #34360 (2026-07-20): Pro account on the Codex app 26.715.52143 blocked at 99% remaining; a banked weekly reset only produced a new "try again at" time — https://github.com/openai/codex/issues/34360 (verified 2026-09-17) 2. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 4. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 5. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 6. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 7. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### 「You've hit your usage limit. Upgrade your plan or add credits to continue, or try again at {Mon D, YYYY, H:MM AM}.」 — Codex URL: https://didreset.com/zh/errors/codex-you-have-hit-your-usage-limit-upgrade-or-add-credits-try-again-at Language: zh · Surface: App · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade your plan or add credits to continue, or try again at {Mon D, YYYY, H:MM AM}. 「You've hit your usage limit. Upgrade your plan or add credits…」是 Codex 在你的 5 小时窗口或周限额之一用完时的提示。它按你账号的时钟刷新，/status 能看到时间；不是本站板面追踪的全员重置。有记录的最近一次 Codex 官方重置：{lastResetDate}。 「You've hit your usage limit. Upgrade your plan or add credits to continue, or try again at {Mon D, YYYY, H:MM AM}.」是什么意思？ 这是 2026 年中以后个人版 ChatGPT 计划里最常见的 Codex 限额拦截文案，桌面 App、CLI、IDE 扩展都是同一句式。OpenAI 按两个窗口计量 Codex：「a five-hour window and a weekly window」，「try again at」后面的时间就是用尽的那个窗口在你账号上的刷新时间。 句子里的三个选项与 OpenAI 帮助中心一致：「You may be able to add credits, use an available reset, upgrade, or wait until the displayed reset time.」符合条件的 Plus 和 Pro 用户可以不换计划直接购买 credits。 原文出处是 openai/codex issue #34360（2026-07-20）：Pro 用户在 Windows 版 Codex App 26.715.52143 上被拦，App 同时显示剩余 99%，兑换一张周重置卡后只换来一个新时间戳（Jul 27, 2026, 10:26 AM）。OpenAI 给它打了 bug / rate-limits 标签，也就是说这句话可能在你的进度条还有余量时出现。 满的是哪个窗口？ - 窗口: 5 小时窗口 · 何时刷新: 从打开窗口的第一条消息起 5 小时，按账号计算。OpenAI 原话：「You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining.」2026-07-12 对 Plus、Business、Pro 暂时取消；Plus 自 2026-08-25 恢复（原帖写「Tomorrow」，帖子时间 2026-08-25 01:16 UTC，即太平洋时间 8 月 25 日）；Pro $100 / $200「for the upcoming months」不启用。 · 在哪里看: 在 Codex CLI 会话里输入 /status（「5h limit」行）；App 的 Settings → Usage；chatgpt.com/codex/settings/usage - 窗口: 周限额 · 何时刷新: 每周一次，按你账号自己的日期。OpenAI 没有公布周限额数字（原话「Weekly limits may also apply」）。兑换整张重置卡会「changes your weekly reset date」；购买即时重置后，下一次自动周刷新是你第一条请求之后的第 7 天。 · 在哪里看: /status（「Weekly limit」行）；Settings → Usage；chatgpt.com/codex/settings/usage 看到它该怎么办 1. 先确认满的是哪个窗口: 在 Codex CLI 会话里输入 /status，或打开 App 的 Settings → Usage。OpenAI 帮助中心原话：「Check which allowance is exhausted, your credit balance, and any reset time shown.」「5h limit」和「Weekly limit」两行各有自己的刷新时间。 2. 看看有没有重置卡: 个人菜单显示「1 reset available」时，兑换整张重置卡会「refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date」。只有真的刷新了至少一个窗口，卡才会被消耗。 3. 加购 credits、升级，或者等: 符合条件的 Plus 和 Pro 用户可以不换计划购买 credits；credits 是付费加量，不是重置。不想花钱就等到文案里的时间。 4. 数字对不上就报告: 如果 /status 或 Settings → Usage 显示两个窗口都还有余量却被拦，带 /status 输出去 github.com/openai/codex 提 issue（标签 rate-limits），或联系 OpenAI Support，附上显示的重置时间、截图、客户端与模型、以及你的时区。 什么时候重置？ 你自己的窗口：「try again at」后面的时间就是用尽的那个窗口在你账号上的刷新时间。5 小时窗口从打开它的那条消息起 5 小时后滚动，周限额按你账号自己的周日期刷新。把这个时间贴进窗口倒计时，就能得到倒计时和日历文件。 全员重置：OpenAI 偶尔会宣布面向所有人的重置，「An automatic reset is applied directly to the usage limits covered by the announcement.」这类事件就是出发板和 Codex 历史页记录的内容。它们没有排期，所以板上只显示最近一次和基于间隔的估计。 FAQ: Q: 「try again at」指的是哪个窗口？ A: 用完的那个。Codex 按账号同时记 5 小时窗口和周限额，这句话本身不说是哪一个，但 /status 会在「5h limit」和「Weekly limit」旁边各打一个刷新时间。几小时后的时间通常是 5 小时窗口，几天后的日期则是周限额。 Q: 我用了重置卡还是看到这句话，为什么？ A: 这正是 issue #34360 报告的 bug：Pro 用户剩余 99% 时用掉一张周重置卡，只换来一个新的「try again at」时间。OpenAI 说重置卡「is consumed only when it successfully refreshes at least one eligible usage window」，所以先到 Settings → Usage 看刷新时间有没有变，卡被消耗却什么都没刷新就去报告。 Q: OpenAI Support 能帮我重置额度吗？ A: 不能。帮助中心写得很明确：「OpenAI Support does not reset ChatGPT or Codex usage limits. If you reach a limit, wait until it resets or use another available option shown in your account.」只有用量被算错、或过了显示的重置时间仍无法使用时，Support 才会介入调查。 Q: 全员重置会清掉这条提示吗？ A: OpenAI 宣布全员重置时会：自动重置「is applied directly to the usage limits covered by the announcement」，不需要你手动兑换。上方板面行显示最近 24 小时内有没有这样的重置；它不能预测下一次。 Sources: 1. openai/codex issue #34360 (2026-07-20): Pro account on the Codex app 26.715.52143 blocked at 99% remaining; a banked weekly reset only produced a new "try again at" time — https://github.com/openai/codex/issues/34360 (verified 2026-09-17) 2. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 4. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 5. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 6. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 7. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### "You've hit your usage limit. Upgrade to Plus to continue using Codex (https://chatgpt.com/explore/plus), or try again at {Mon Dth, YYYY H:MM AM}." — Codex URL: https://didreset.com/errors/codex-you-have-hit-your-usage-limit-upgrade-to-plus-try-again-at Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade to Plus to continue using Codex (https://chatgpt.com/explore/plus), or try again at {Mon Dth, YYYY H:MM AM}. "You've hit your usage limit. Upgrade to Plus to continue using Codex (https://chatgpt.com/explore/plus), or try again at {Mon Dth, YYYY H:MM AM}." is the Codex limit notice shown to accounts without a Plus entitlement, and, in reported bugs, to Plus and Business accounts too. Your window refreshes on its own clock, shown by /status; it is not the account-wide reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "You've hit your usage limit. Upgrade to Plus to continue using Codex (https://chatgpt.com/explore/plus), or try again at {Mon Dth, YYYY H:MM AM}." mean? This is the "Upgrade to Plus" variant of the Codex usage-limit message: the same two-window metering (5-hour window and weekly limit) as every other Codex limit notice, but the upsell points at Plus because the account that hit it has no Plus entitlement, or Codex thinks it does not. Both source reports were wrong about the plan. In openai/codex issue #30349 (2026-06-27, CLI 0.142.3) a Plus account saw this sentence while /status printed "5h limit … 99% left" and "Weekly limit … 100% left"; in #29948 (2026-06-25) a Business account got the identical "Upgrade to Plus" text. OpenAI labelled both bug / rate-limits. So when the plan in the sentence does not match the plan you pay for, treat the "try again at" time as unreliable and check the account line in /status first. When the plan does match, the sentence simply means your weekly or 5-hour allowance is used up until the time shown. Which window is full? - Window: 5-hour window · Resets: Five hours after the first message that opened the window, per account. OpenAI: "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." Removed for Plus, Business and Pro on 2026-07-12; back for Plus since 2026-08-25 (the post says "Tomorrow" and is stamped 2026-08-25 01:16 UTC, i.e. Pacific time 2026-08-25); "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. · Where to check: /status inside an active Codex CLI session (the "5h limit" line); Settings → Usage in the app; chatgpt.com/codex/settings/usage - Window: Weekly limit · Resets: Once a week on your account's own date. OpenAI publishes no weekly figure ("Weekly limits may also apply"). A full banked reset "changes your weekly reset date"; after a purchased reset the next automatic weekly reset is seven days after your first request. · Where to check: /status (the "Weekly limit" line); Settings → Usage; chatgpt.com/codex/settings/usage What to do when you see it 1. Run /status and read the Account line: In the Codex session type /status. The box shows "Account:" with the plan Codex believes you have, plus the "5h limit" and "Weekly limit" bars with their reset times. If it says Plus or Business and both bars have room, the message is the known bug. 2. Sign out and sign in again: Type /logout ("Sign out of Codex. Clear local credentials"), then run codex login and complete the browser flow; codex login --device-auth is the alternative used in issue #30349. Codex caches credentials in ~/.codex/auth.json or your OS credential store, so a fresh login replaces a stale entitlement. 3. Report a plan mismatch: If a Plus or Business account still gets "Upgrade to Plus", open an issue at github.com/openai/codex with a screenshot of /status, or contact OpenAI Support with the reset time shown, the client and model, and your time zone. 4. Otherwise wait for the time in the message: On a Free or Go account the sentence is accurate: the exhausted window refreshes at the "try again at" time, or you upgrade. OpenAI Support does not reset Codex usage limits. When does it reset? Your own window: if the plan in the message is correct, the "try again at" time is when your exhausted window refreshes; a date days away means the weekly limit, a time hours away means the 5-hour window. Paste it into the window tracker for a countdown. If the plan is wrong, the timestamp in the bug reports was not trustworthy; re-login first. Account-wide resets: OpenAI's occasional resets for everyone are applied automatically to "the usage limits covered by the announcement". The board row above shows the last one on record and a cadence estimate; the Codex history page lists every announcement with its original post. FAQ: Q: I am on Plus (or Business). Why does it tell me to upgrade to Plus? A: Because Codex has mis-read your entitlement. That exact situation is documented in openai/codex issues #30349 (Plus, /status showing 99% and 100% left) and #29948 (Business), both labelled bug / rate-limits by OpenAI. Sign out with /logout, sign in again with codex login, and report it if the sentence comes back. Q: Is the "try again at" time reliable in this message? A: Only when the plan it names matches yours. In the bug reports the timestamp appeared while both usage bars were nearly full, so it did not describe a real window. On a Free or Go account the time is your normal window refresh. Q: Does logging out and back in reset my usage? A: No. /logout only clears local credentials ("Clear local credentials when using a shared machine"), and the allowance belongs to your plan: "Local messages and cloud chats share your plan's usage allowance." Re-login fixes a stale entitlement, not an empty window. Q: Would an account-wide reset fix this? A: Only if your window is genuinely empty. An OpenAI-announced reset refreshes "the usage limits covered by the announcement" for everyone; it does not correct a plan mismatch. Check the board row above for the last reset on record. Sources: 1. openai/codex issue #30349 (2026-06-27): Plus account on Codex CLI 0.142.3 blocked with "Upgrade to Plus" while /status showed 5h 99% left and Weekly 100% left — https://github.com/openai/codex/issues/30349 (verified 2026-09-17) 2. openai/codex issue #29948 (2026-06-25): Business account on Codex CLI 0.142.0 shown the same "Upgrade to Plus" sentence (try again at Jul 19th, 2026 10:27 AM); labelled bug / rate-limits — https://github.com/openai/codex/issues/29948 (verified 2026-09-17) 3. ChatGPT Learn: Authentication (codex login, codex login status, codex logout, login caching) — https://learn.chatgpt.com/docs/auth (verified 2026-09-18) 4. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 5. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 6. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 7. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 8. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 9. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 10. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### 「You've hit your usage limit. Upgrade to Plus to continue using Codex (https://chatgpt.com/explore/plus), or try again at {Mon Dth, YYYY H:MM AM}.」 — Codex URL: https://didreset.com/zh/errors/codex-you-have-hit-your-usage-limit-upgrade-to-plus-try-again-at Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade to Plus to continue using Codex (https://chatgpt.com/explore/plus), or try again at {Mon Dth, YYYY H:MM AM}. 「You've hit your usage limit. Upgrade to Plus…」是 Codex 对没有 Plus 权益的账号显示的限额提示；已报告的 bug 里 Plus 和 Business 账号也会看到。窗口按自己的时钟刷新，/status 能看到时间；不是本站板面追踪的全员重置。最近一次官方重置：{lastResetDate}。 「You've hit your usage limit. Upgrade to Plus to continue using Codex (https://chatgpt.com/explore/plus), or try again at {Mon Dth, YYYY H:MM AM}.」是什么意思？ 这是 Codex 限额文案的「Upgrade to Plus」版本：计量方式和其他版本一样（5 小时窗口 + 周限额），但推销的是 Plus，因为触发它的账号没有 Plus 权益，或者 Codex 以为没有。 两条出处的账号计划都对不上。openai/codex issue #30349（2026-06-27，CLI 0.142.3）里，Plus 账号看到这句话时 /status 打印的是「5h limit … 99% left」和「Weekly limit … 100% left」；#29948（2026-06-25）里 Business 账号也收到一模一样的「Upgrade to Plus」。OpenAI 给两条都打了 bug / rate-limits 标签。 所以当句子里的计划和你付费的计划不符时，把「try again at」的时间当作不可信，先看 /status 的 Account 行。计划相符时，这句话就是字面意思：周限额或 5 小时窗口在显示的时间之前已用尽。 满的是哪个窗口？ - 窗口: 5 小时窗口 · 何时刷新: 从打开窗口的第一条消息起 5 小时，按账号计算。OpenAI 原话：「You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining.」2026-07-12 对 Plus、Business、Pro 暂时取消；Plus 自 2026-08-25 恢复（原帖写「Tomorrow」，帖子时间 2026-08-25 01:16 UTC，即太平洋时间 8 月 25 日）；Pro $100 / $200「for the upcoming months」不启用。 · 在哪里看: 在 Codex CLI 会话里输入 /status（「5h limit」行）；App 的 Settings → Usage；chatgpt.com/codex/settings/usage - 窗口: 周限额 · 何时刷新: 每周一次，按你账号自己的日期。OpenAI 没有公布周限额数字（原话「Weekly limits may also apply」）。兑换整张重置卡会「changes your weekly reset date」；购买即时重置后，下一次自动周刷新是你第一条请求之后的第 7 天。 · 在哪里看: /status（「Weekly limit」行）；Settings → Usage；chatgpt.com/codex/settings/usage 看到它该怎么办 1. 输入 /status，先看 Account 行: 在 Codex 会话里输入 /status。框里的「Account:」是 Codex 认为你拥有的计划，下面是「5h limit」和「Weekly limit」两条进度条和各自的刷新时间。如果显示 Plus 或 Business 且两条都有余量，这就是已知 bug。 2. 退出登录再登录: 输入 /logout（「Sign out of Codex. Clear local credentials」），再运行 codex login 完成浏览器授权；issue #30349 里也试过 codex login --device-auth。Codex 把凭据缓存在 ~/.codex/auth.json 或系统凭据库，重新登录会替换掉过期的权益信息。 3. 计划对不上就报告: Plus 或 Business 账号仍然看到「Upgrade to Plus」，就带 /status 截图去 github.com/openai/codex 提 issue，或联系 OpenAI Support，附上显示的重置时间、客户端与模型、以及你的时区。 4. 否则等到文案里的时间: Free 或 Go 账号看到这句话是正常的：用尽的窗口会在「try again at」的时间刷新，或者升级。OpenAI Support 不会重置 Codex 额度。 什么时候重置？ 你自己的窗口：如果文案里的计划没错，「try again at」就是用尽窗口的刷新时间；几天后的日期是周限额，几小时后的时间是 5 小时窗口。贴进窗口倒计时即可。如果计划错了，bug 报告里的时间戳并不可信，先重新登录。 全员重置：OpenAI 偶尔宣布的全员重置会自动应用到「the usage limits covered by the announcement」。上方板面行显示有记录的最近一次和基于间隔的估计；Codex 历史页列出每一次公告和原帖。 FAQ: Q: 我是 Plus（或 Business），为什么让我升级 Plus？ A: 因为 Codex 读错了你的权益。这正是 openai/codex issue #30349（Plus，/status 显示 99% 和 100% 剩余）和 #29948（Business）记录的情况，OpenAI 都打了 bug / rate-limits 标签。先 /logout 再 codex login 重新登录，再出现就去报告。 Q: 这条文案里的「try again at」时间可信吗？ A: 只有它说的计划和你的一致时才可信。bug 报告里的时间戳出现在两条进度条都几乎满格的时候，并不对应真实的窗口。Free 或 Go 账号看到的时间就是正常的窗口刷新时间。 Q: 退出再登录会重置我的额度吗？ A: 不会。/logout 只清除本地凭据（「Clear local credentials when using a shared machine」），额度属于计划本身：「Local messages and cloud chats share your plan's usage allowance.」重新登录修的是过期的权益信息，不是空掉的窗口。 Q: 全员重置能解决这个问题吗？ A: 只有在你的窗口真的用完时才有用。OpenAI 宣布的重置会为所有人刷新「the usage limits covered by the announcement」，但不会纠正计划识别错误。上方板面行显示有记录的最近一次重置。 Sources: 1. openai/codex issue #30349 (2026-06-27): Plus account on Codex CLI 0.142.3 blocked with "Upgrade to Plus" while /status showed 5h 99% left and Weekly 100% left — https://github.com/openai/codex/issues/30349 (verified 2026-09-17) 2. openai/codex issue #29948 (2026-06-25): Business account on Codex CLI 0.142.0 shown the same "Upgrade to Plus" sentence (try again at Jul 19th, 2026 10:27 AM); labelled bug / rate-limits — https://github.com/openai/codex/issues/29948 (verified 2026-09-17) 3. ChatGPT Learn: Authentication (codex login, codex login status, codex logout, login caching) — https://learn.chatgpt.com/docs/auth (verified 2026-09-18) 4. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 5. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 6. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 7. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 8. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 9. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 10. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### "You've hit your usage limit. Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits or try again at {Mon Dth, YYYY H:MM PM}." — Codex URL: https://didreset.com/errors/codex-you-have-hit-your-usage-limit-upgrade-to-pro-purchase-more-credits Language: en · Surface: IDE · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits or try again at {Mon Dth, YYYY H:MM PM}. "You've hit your usage limit. Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits or try again at {Mon Dth, YYYY H:MM PM}." is Codex's message to Plus accounts whose weekly limit or 5-hour window is used up. It refreshes on your account's own clock, shown by /status and the usage page; it is not the account-wide reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "You've hit your usage limit. Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits or try again at {Mon Dth, YYYY H:MM PM}." mean? This is the Plus-plan variant of the Codex limit notice, seen in the VS Code extension: instead of a generic "upgrade your plan" it offers the two paid routes OpenAI gives Plus users, upgrading to Pro or buying credits on the usage page, and the "try again at" time is when the exhausted window refreshes. The source report (openai/codex issue #12299, 2026-02-20, extension 0.4.76 on Linux) was filed with a "try again at Feb 23rd, 2026 9:01 PM" timestamp three days out, which points at the weekly limit, and the reporter still had 10% of the weekly allowance showing when it appeared. OpenAI describes the same options in its help center: "You may be able to add credits, use an available reset, upgrade, or wait until the displayed reset time. Eligible Plus and Pro users can buy credits without changing plans." Credits are extra usage under their own terms; "A reset is not cash or API credit." Which window is full? - Window: 5-hour window · Resets: Five hours after the first message that opened the window, per account. OpenAI: "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." Removed for Plus, Business and Pro on 2026-07-12; back for Plus since 2026-08-25 (the post says "Tomorrow" and is stamped 2026-08-25 01:16 UTC, i.e. Pacific time 2026-08-25); "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. · Where to check: /status inside an active Codex CLI session (the "5h limit" line); Settings → Usage in the app; chatgpt.com/codex/settings/usage - Window: Weekly limit · Resets: Once a week on your account's own date. OpenAI publishes no weekly figure ("Weekly limits may also apply"). A full banked reset "changes your weekly reset date"; after a purchased reset the next automatic weekly reset is seven days after your first request. · Where to check: /status (the "Weekly limit" line); Settings → Usage; chatgpt.com/codex/settings/usage What to do when you see it 1. Open the usage page: Go to https://chatgpt.com/codex/settings/usage (the URL in the message) or, in a CLI session, type /status. Both show the 5-hour and weekly bars with their reset times; the help center says to "Check which allowance is exhausted, your credit balance, and any reset time shown." 2. Decide between credits, Pro, a reset, or waiting: Buying credits keeps your Plus plan and pays for additional usage. Upgrading to Pro raises the included allowance. A banked reset, if you have one, refreshes both windows. Waiting costs nothing: the window refreshes at the time in the message. 3. Report a mismatch: If the usage page still shows remaining allowance, as in issue #12299 at 10% weekly, file an issue at github.com/openai/codex with a screenshot, or contact OpenAI Support with the reset time shown, the client and model, and your time zone. When does it reset? Your own window: the "try again at" time is your account's refresh for the window that ran out. A date days away is the weekly limit; a time within a few hours is the 5-hour window. Paste it into the window tracker for a countdown and a calendar file. Account-wide resets: OpenAI occasionally announces a reset for everyone, applied automatically to "the usage limits covered by the announcement". The board row above shows the last one on record; the Codex history page lists every announcement with its original post. They are not scheduled and cannot be predicted from this message. FAQ: Q: Is buying credits the same as a reset? A: No. OpenAI's help center: "Credits pay for eligible additional usage under their own terms. A reset is not cash or API credit." Credits let you keep working past the limit; a reset refreshes the windows themselves. Q: Can I buy a reset instead of credits? A: OpenAI says "Purchased instant resets are available to eligible personal ChatGPT Plus and Pro accounts, depending on your account and billing country." A purchased reset refreshes both windows as soon as checkout succeeds and cannot be saved for later; the help center does not publish a price. Q: Why did it fire with 10% of my weekly allowance left? A: That is what issue #12299 reports, and OpenAI has not published how the cut-off is computed. If your usage page still shows allowance, report it; the help center asks you to include the reset time shown, a screenshot, the Codex client and model, and your time zone. Q: Does Pro have the same windows? A: Pro keeps the weekly limit, but on 2026-08-25 Tibo Sottiaux said the 5-hour limit is "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months, while it came back for Plus. OpenAI publishes five-hour estimates per plan but no weekly figures. Sources: 1. openai/codex issue #12299 (2026-02-20): Plus user on the Codex VS Code extension 0.4.76 (Linux) blocked with 10% of the weekly allowance still showing; try again at Feb 23rd, 2026 9:01 PM — https://github.com/openai/codex/issues/12299 (verified 2026-09-17) 2. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 4. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 5. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 6. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 7. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### 「You've hit your usage limit. Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits or try again at {Mon Dth, YYYY H:MM PM}.」 — Codex URL: https://didreset.com/zh/errors/codex-you-have-hit-your-usage-limit-upgrade-to-pro-purchase-more-credits Language: zh · Surface: IDE · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits or try again at {Mon Dth, YYYY H:MM PM}. 「You've hit your usage limit. Upgrade to Pro…purchase more credits…」是 Codex 对周限额或 5 小时窗口用尽的 Plus 账号的提示。窗口按你账号的时钟刷新，/status 能看到时间；不是本站板面追踪的全员重置。最近一次 Codex 官方重置：{lastResetDate}。 「You've hit your usage limit. Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits or try again at {Mon Dth, YYYY H:MM PM}.」是什么意思？ 这是 Codex 限额提示的 Plus 版本，出现在 VS Code 扩展里：不再泛泛地说「升级计划」，而是给出 OpenAI 提供给 Plus 用户的两条付费路线，升 Pro 或者到用量页购买 credits；「try again at」后面是用尽窗口的刷新时间。 出处 openai/codex issue #12299（2026-02-20，Linux 上的扩展 0.4.76）里的时间戳是「try again at Feb 23rd, 2026 9:01 PM」，三天以后，指向周限额；而报告者的周额度当时还显示剩 10%。 OpenAI 帮助中心列的选项与此一致：「You may be able to add credits, use an available reset, upgrade, or wait until the displayed reset time. Eligible Plus and Pro users can buy credits without changing plans.」credits 是按自己条款计费的额外用量，「A reset is not cash or API credit.」 满的是哪个窗口？ - 窗口: 5 小时窗口 · 何时刷新: 从打开窗口的第一条消息起 5 小时，按账号计算。OpenAI 原话：「You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining.」2026-07-12 对 Plus、Business、Pro 暂时取消；Plus 自 2026-08-25 恢复（原帖写「Tomorrow」，帖子时间 2026-08-25 01:16 UTC，即太平洋时间 8 月 25 日）；Pro $100 / $200「for the upcoming months」不启用。 · 在哪里看: 在 Codex CLI 会话里输入 /status（「5h limit」行）；App 的 Settings → Usage；chatgpt.com/codex/settings/usage - 窗口: 周限额 · 何时刷新: 每周一次，按你账号自己的日期。OpenAI 没有公布周限额数字（原话「Weekly limits may also apply」）。兑换整张重置卡会「changes your weekly reset date」；购买即时重置后，下一次自动周刷新是你第一条请求之后的第 7 天。 · 在哪里看: /status（「Weekly limit」行）；Settings → Usage；chatgpt.com/codex/settings/usage 看到它该怎么办 1. 打开用量页: 访问文案里的 https://chatgpt.com/codex/settings/usage，或在 CLI 会话里输入 /status。两处都显示 5 小时与周限额进度条和刷新时间；帮助中心原话：「Check which allowance is exhausted, your credit balance, and any reset time shown.」 2. 在 credits、Pro、重置卡和等待之间选: 买 credits 可以保留 Plus 计划、按量付费继续用；升 Pro 提高包含的额度；有重置卡就兑换，两个窗口一起刷新；不花钱就等到文案里的时间。 3. 数字对不上就报告: 如果用量页还显示有余量（像 issue #12299 里周额度剩 10%），带截图去 github.com/openai/codex 提 issue，或联系 OpenAI Support，附上显示的重置时间、客户端与模型、以及你的时区。 什么时候重置？ 你自己的窗口：「try again at」就是用尽窗口在你账号上的刷新时间。几天后的日期是周限额，几小时内的时间是 5 小时窗口。贴进窗口倒计时就能得到倒计时和日历文件。 全员重置：OpenAI 偶尔宣布面向所有人的重置，会自动应用到「the usage limits covered by the announcement」。上方板面行显示有记录的最近一次；Codex 历史页列出每一次公告和原帖。它们没有排期，也无法从这句话推断。 FAQ: Q: 买 credits 等于重置吗？ A: 不等于。OpenAI 帮助中心原话：「Credits pay for eligible additional usage under their own terms. A reset is not cash or API credit.」credits 让你在超限后继续用，重置则是把窗口本身刷新。 Q: 能不买 credits、直接买一次重置吗？ A: OpenAI 说「Purchased instant resets are available to eligible personal ChatGPT Plus and Pro accounts, depending on your account and billing country.」购买的重置在付款成功后立即刷新两个窗口，不能存起来；帮助中心没有公布价格。 Q: 周额度还剩 10% 为什么就被拦了？ A: 这是 issue #12299 报告的现象，OpenAI 没有公布截断是怎么算的。用量页仍显示有余量就去报告；帮助中心要求附上显示的重置时间、截图、Codex 客户端与模型、以及你的时区。 Q: Pro 的窗口一样吗？ A: Pro 保留周限额，但 Tibo Sottiaux 在 2026-08-25 说 5 小时限制「not enabled for Pro $100 and Pro $200 subscriptions」，未来几个月都不启用，而 Plus 已恢复。OpenAI 公布了各计划的 5 小时估算，没有公布周限额数字。 Sources: 1. openai/codex issue #12299 (2026-02-20): Plus user on the Codex VS Code extension 0.4.76 (Linux) blocked with 10% of the weekly allowance still showing; try again at Feb 23rd, 2026 9:01 PM — https://github.com/openai/codex/issues/12299 (verified 2026-09-17) 2. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 4. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 5. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 6. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 7. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### "You've hit your usage limit. To get more access now, send a request to your admin or try again at {Mon Dth, YYYY H:MM PM}." — Codex URL: https://didreset.com/errors/codex-you-have-hit-your-usage-limit-send-a-request-to-your-admin Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. To get more access now, send a request to your admin or try again at {Mon Dth, YYYY H:MM PM}. "You've hit your usage limit. To get more access now, send a request to your admin or try again at {Mon Dth, YYYY H:MM PM}." is the Codex limit notice for Business and Team workspace seats, where the allowance is controlled by an administrator rather than by upgrading. It refreshes on the workspace's clock, shown by /status; it is not the account-wide reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "You've hit your usage limit. To get more access now, send a request to your admin or try again at {Mon Dth, YYYY H:MM PM}." mean? This is the workspace variant of the Codex usage-limit message. Instead of "upgrade" it tells you to ask your admin, because on Business and Team plans seat allowances and workspace spend caps are set in the workspace, not by the member. The "try again at" time is when the exhausted window refreshes. All three source reports were filed on 2026-04-06 by Business seats (openai/codex issues #16917, #16909, #16906). In #16917 the reporter had used nothing in the previous 30 days, the web app showed "You've reached your workspace spend cap", and /status printed "limits: data not available yet", which points at a workspace cap or a limit-data sync problem rather than a genuinely empty window. OpenAI's help center says Codex, ChatGPT Work and Workspace Agents "use a shared allowance and credit pool when those features are available on your plan"; on a workspace plan that pool and any spend cap are what the admin controls. Which window is full? - Window: 5-hour window · Resets: Five hours after the first message that opened the window, per account. OpenAI: "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." Removed for Plus, Business and Pro on 2026-07-12; back for Plus since 2026-08-25 (the post says "Tomorrow" and is stamped 2026-08-25 01:16 UTC, i.e. Pacific time 2026-08-25); "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. · Where to check: /status inside an active Codex CLI session (the "5h limit" line); Settings → Usage in the app; chatgpt.com/codex/settings/usage - Window: Weekly limit · Resets: Once a week on your account's own date. OpenAI publishes no weekly figure ("Weekly limits may also apply"). A full banked reset "changes your weekly reset date"; after a purchased reset the next automatic weekly reset is seven days after your first request. · Where to check: /status (the "Weekly limit" line); Settings → Usage; chatgpt.com/codex/settings/usage - Window: Workspace spend cap (Business / Team) · Resets: Set and changed by your workspace admin; OpenAI publishes no figure. In issue #16917 the web app showed "You've reached your workspace spend cap" alongside this message. · Where to check: Ask your admin; the banner in the Codex web app; /status may show "limits: data not available yet" while the cap is in force What to do when you see it 1. Ask your workspace admin: The message is literal: on Business and Team plans a member cannot raise the allowance. Ask the admin to check the seat allowance and the workspace spend cap in Codex settings, and to confirm the workspace still has credits. 2. Check whether limit data is missing: Type /status in the session. If it prints "limits: data not available yet", as in issue #16917, the client has no limit data to show; wait a while or sign out with /logout and sign in again with codex login before assuming the window is empty. 3. Report it with workspace details: If the workspace has allowance and you still see the message, file an issue at github.com/openai/codex with the /status output, or have the admin contact OpenAI Support with the reset time shown, a screenshot, the client and model, and your time zone. When does it reset? Your own window: when the message is about a genuine window, the "try again at" time is the refresh time for your seat; a date days away is the weekly limit. Paste it into the window tracker for a countdown. When it is a workspace spend cap, the time in the message may not help: access returns when the admin raises the cap or the workspace period rolls over. Account-wide resets: OpenAI occasionally announces a reset for everyone, applied automatically to "the usage limits covered by the announcement". Whether a specific workspace cap is covered is stated only in that announcement. The board row above shows the last reset on record; the Codex history page lists every announcement with its original post. FAQ: Q: Why does /status say "limits: data not available yet"? A: OpenAI has not documented that line. In issue #16917 it appeared together with this message and a workspace spend-cap banner on a seat with zero usage in the previous 30 days, so it indicated the client had no limit data rather than a used-up window. Re-login and ask the admin to check the workspace settings. Q: Can I buy credits or upgrade myself on a Business seat? A: The message does not offer that route: it says to send a request to your admin. OpenAI's help center points business users to "Codex credit plans for business users" on chatgpt.com/pricing, which are managed at the workspace level. Q: Will an account-wide Codex reset lift a workspace spend cap? A: Not published. OpenAI says an automatic reset "is applied directly to the usage limits covered by the announcement"; whether a workspace cap is among them depends on that announcement. Check the board row above for the last reset on record and read its source post. Q: Can OpenAI Support raise my limit? A: No. "OpenAI Support does not reset ChatGPT or Codex usage limits." Support will investigate if usage was counted incorrectly or access has not returned after the stated reset time, and admins manage the workspace cap. Sources: 1. openai/codex issue #16917 (2026-04-06): Business seat on codex-cli 0.118.0 blocked with zero usage in the past 30 days; the web app showed "You've reached your workspace spend cap" and /status showed "limits: data not available yet" — https://github.com/openai/codex/issues/16917 (verified 2026-09-17) 2. openai/codex issue #16909 (2026-04-06): same-day Business workspace report of the same message — https://github.com/openai/codex/issues/16909 (verified 2026-09-17) 3. openai/codex issue #16906 (2026-04-06): same-day Business workspace report of the same message — https://github.com/openai/codex/issues/16906 (verified 2026-09-17) 4. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 5. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 6. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 7. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 8. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 9. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 10. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### 「You've hit your usage limit. To get more access now, send a request to your admin or try again at {Mon Dth, YYYY H:MM PM}.」 — Codex URL: https://didreset.com/zh/errors/codex-you-have-hit-your-usage-limit-send-a-request-to-your-admin Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. To get more access now, send a request to your admin or try again at {Mon Dth, YYYY H:MM PM}. 「You've hit your usage limit…send a request to your admin…」是 Codex 对 Business / Team 工作区座席的限额提示：额度由管理员控制。它按工作区的时钟刷新，/status 能看到时间；不是本站板面追踪的全员重置。最近一次 Codex 官方重置：{lastResetDate}。 「You've hit your usage limit. To get more access now, send a request to your admin or try again at {Mon Dth, YYYY H:MM PM}.」是什么意思？ 这是 Codex 限额文案的工作区版本。它不说「升级」，而是让你向 admin 申请，因为 Business 和 Team 计划的座席额度与工作区 spend cap 是在工作区里设置的，成员自己改不了。「try again at」后面是用尽窗口的刷新时间。 三条出处都是 2026-04-06 同一天由 Business 座席提交的（openai/codex issues #16917、#16909、#16906）。#16917 的报告者过去 30 天用量为 0，网页版 Codex 显示「You've reached your workspace spend cap」，/status 打印「limits: data not available yet」，指向工作区上限或限额数据未同步，而不是窗口真的用完。 OpenAI 帮助中心说 Codex、ChatGPT Work 和 Workspace Agents「use a shared allowance and credit pool when those features are available on your plan」；在工作区计划里，这个池子和 spend cap 都由管理员掌控。 满的是哪个窗口？ - 窗口: 5 小时窗口 · 何时刷新: 从打开窗口的第一条消息起 5 小时，按账号计算。OpenAI 原话：「You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining.」2026-07-12 对 Plus、Business、Pro 暂时取消；Plus 自 2026-08-25 恢复（原帖写「Tomorrow」，帖子时间 2026-08-25 01:16 UTC，即太平洋时间 8 月 25 日）；Pro $100 / $200「for the upcoming months」不启用。 · 在哪里看: 在 Codex CLI 会话里输入 /status（「5h limit」行）；App 的 Settings → Usage；chatgpt.com/codex/settings/usage - 窗口: 周限额 · 何时刷新: 每周一次，按你账号自己的日期。OpenAI 没有公布周限额数字（原话「Weekly limits may also apply」）。兑换整张重置卡会「changes your weekly reset date」；购买即时重置后，下一次自动周刷新是你第一条请求之后的第 7 天。 · 在哪里看: /status（「Weekly limit」行）；Settings → Usage；chatgpt.com/codex/settings/usage - 窗口: 工作区 spend cap（Business / Team） · 何时刷新: 由工作区管理员设置和调整；OpenAI 未公布数字。issue #16917 里网页版与这条文案同时显示「You've reached your workspace spend cap」。 · 在哪里看: 问管理员；Codex 网页版的横幅；cap 生效期间 /status 可能显示「limits: data not available yet」 看到它该怎么办 1. 找工作区管理员: 这句话就是字面意思：Business 和 Team 计划的成员无法自己提高额度。请管理员在 Codex 设置里检查座席额度和工作区 spend cap，并确认工作区还有 credits。 2. 确认限额数据是不是缺失: 在会话里输入 /status。如果打印「limits: data not available yet」（issue #16917 的情况），说明客户端拿不到限额数据；先等一会儿，或者 /logout 后重新 codex login，再判断窗口是否真的用完。 3. 带工作区信息去报告: 工作区有额度却仍被拦，就带 /status 输出去 github.com/openai/codex 提 issue，或请管理员联系 OpenAI Support，附上显示的重置时间、截图、客户端与模型、以及时区。 什么时候重置？ 你自己的窗口：文案针对的是真实窗口时，「try again at」就是你这个座席的刷新时间；几天后的日期是周限额。贴进窗口倒计时即可。如果是工作区 spend cap，文案里的时间未必有用：管理员调高上限或工作区周期滚动后才恢复。 全员重置：OpenAI 偶尔宣布面向所有人的重置，会自动应用到「the usage limits covered by the announcement」；具体某个工作区上限是否包含在内，只有那次公告本身会说。上方板面行显示有记录的最近一次；Codex 历史页列出每一次公告和原帖。 FAQ: Q: /status 为什么显示「limits: data not available yet」？ A: OpenAI 没有解释这一行。issue #16917 里它和这条文案、工作区 spend cap 横幅一起出现在一个过去 30 天零用量的座席上，说明客户端没拿到限额数据，而不是窗口用完了。重新登录，并请管理员检查工作区设置。 Q: Business 座席能自己买 credits 或升级吗？ A: 这条文案没有给这个选项：它让你向 admin 申请。OpenAI 帮助中心把企业用户引导到 chatgpt.com/pricing 的「Codex credit plans for business users」，那是在工作区层面管理的。 Q: 全员重置会解除工作区 spend cap 吗？ A: 未公布。OpenAI 说自动重置「is applied directly to the usage limits covered by the announcement」，工作区上限是否在内取决于那次公告。看上方板面行有记录的最近一次重置，并读它的原帖。 Q: OpenAI Support 能帮我提高额度吗？ A: 不能。「OpenAI Support does not reset ChatGPT or Codex usage limits.」只有用量被算错或过了显示的重置时间仍无法使用时 Support 才会调查；工作区上限由管理员管理。 Sources: 1. openai/codex issue #16917 (2026-04-06): Business seat on codex-cli 0.118.0 blocked with zero usage in the past 30 days; the web app showed "You've reached your workspace spend cap" and /status showed "limits: data not available yet" — https://github.com/openai/codex/issues/16917 (verified 2026-09-17) 2. openai/codex issue #16909 (2026-04-06): same-day Business workspace report of the same message — https://github.com/openai/codex/issues/16909 (verified 2026-09-17) 3. openai/codex issue #16906 (2026-04-06): same-day Business workspace report of the same message — https://github.com/openai/codex/issues/16906 (verified 2026-09-17) 4. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 5. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 6. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 7. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 8. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 9. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 10. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### "You've hit your usage limit. Upgrade to Pro (https://openai.com/chatgpt/pricing) or try again in {N} days {N} hours {N} minutes." — Codex URL: https://didreset.com/errors/codex-you-have-hit-your-usage-limit-try-again-in-days-hours-minutes Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade to Pro (https://openai.com/chatgpt/pricing) or try again in {N} days {N} hours {N} minutes. "You've hit your usage limit. Upgrade to Pro (https://openai.com/chatgpt/pricing) or try again in {N} days {N} hours {N} minutes." is the 2025 form of Codex's limit notice, with a relative countdown instead of a date. A countdown measured in days means the weekly limit; it refreshes on your account's own clock, shown by /status, and is not the account-wide reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "You've hit your usage limit. Upgrade to Pro (https://openai.com/chatgpt/pricing) or try again in {N} days {N} hours {N} minutes." mean? This wording comes from Codex CLI 0.39.0 in September 2025 (openai/codex issue #3899, 2025-09-19): the same usage-limit block as today, but with the wait expressed as "N days N hours N minutes" rather than an absolute "try again at" time. By 2026 the notice switched to absolute timestamps (issue #34360 shows "try again at Jul 25, 2026, 11:09 AM"), but the old sentence is still what many searches quote. The countdown in the report was "5 days 22 hours 11 minutes". A wait of several days cannot be the 5-hour window, so a message in this form is telling you the weekly limit is used up; OpenAI meters Codex over "a five-hour window and a weekly window" and publishes no weekly figure. The only paid route offered in the 2025 text is upgrading to Pro; newer versions also mention credits and, where offered, a banked or purchased reset. Which window is full? - Window: 5-hour window · Resets: Five hours after the first message that opened the window, per account. OpenAI: "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." Removed for Plus, Business and Pro on 2026-07-12; back for Plus since 2026-08-25 (the post says "Tomorrow" and is stamped 2026-08-25 01:16 UTC, i.e. Pacific time 2026-08-25); "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. · Where to check: /status inside an active Codex CLI session (the "5h limit" line); Settings → Usage in the app; chatgpt.com/codex/settings/usage - Window: Weekly limit · Resets: Once a week on your account's own date. OpenAI publishes no weekly figure ("Weekly limits may also apply"). A full banked reset "changes your weekly reset date"; after a purchased reset the next automatic weekly reset is seven days after your first request. · Where to check: /status (the "Weekly limit" line); Settings → Usage; chatgpt.com/codex/settings/usage What to do when you see it 1. Read the countdown as a weekly-limit wait: Days in the countdown mean the weekly limit is what ran out. Add the countdown to the current time to get your refresh moment, or update the CLI so /status prints the absolute reset time next to "Weekly limit". 2. Check /status in a current CLI: In an active session type /status; OpenAI's help center calls this the way to "Check which allowance is exhausted, your credit balance, and any reset time shown." Settings → Usage in the app and chatgpt.com/codex/settings/usage show the same bars. 3. Upgrade, add credits, redeem a reset, or wait: The 2025 message only offers Pro, but OpenAI now says "You may be able to add credits, use an available reset, upgrade, or wait until the displayed reset time." Waiting is always an option; OpenAI Support does not reset usage limits. When does it reset? Your own window: the countdown is your account's weekly refresh, counted from the moment the message appeared. Paste the resulting time into the window tracker to get a countdown and a calendar file, or run /status in a newer CLI to read the exact time. Account-wide resets: OpenAI occasionally resets everyone's Codex limits and the reset "is applied directly to the usage limits covered by the announcement". Those events are what the board row above and the Codex history page record; they are not scheduled, so a days-long countdown can be shortened only by such an announcement, a banked reset, or a purchased reset. FAQ: Q: Why does it say days instead of a time? A: Because this is the September 2025 wording of the notice, which expressed the wait as "N days N hours N minutes" (issue #3899). Codex CLI releases from 2026 show an absolute time such as "try again at Jul 25, 2026, 11:09 AM" instead (issue #34360). Q: Which window is a five-day countdown? A: The weekly limit. Codex meters a 5-hour window and a weekly window; a wait of "5 days 22 hours 11 minutes" cannot be the five-hour one. OpenAI publishes no weekly figure and says only that "Weekly limits may also apply." Q: Is this message still current? A: The wording is from CLI 0.39.0 (2025-09-19). Later reports of the same block use the absolute "try again at" form, so if you see the countdown version you are on an older CLI; the limit behind it is the same. Q: Does an account-wide reset shorten the countdown? A: Yes, when OpenAI announces one: the automatic reset "is applied directly to the usage limits covered by the announcement". The board row above shows the last such reset on record; it cannot predict the next. Sources: 1. openai/codex issue #3899 (2025-09-19): Codex CLI 0.39.0 (Windows, gpt-5-codex): "try again in 5 days 22 hours 11 minutes" — https://github.com/openai/codex/issues/3899 (verified 2026-09-17) 2. openai/codex issue #34360 (2026-07-20): the 2026 wording of the same notice uses an absolute time: "try again at Jul 25, 2026, 11:09 AM" — https://github.com/openai/codex/issues/34360 (verified 2026-09-17) 3. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 4. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 5. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 6. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 7. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 8. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 9. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### 「You've hit your usage limit. Upgrade to Pro (https://openai.com/chatgpt/pricing) or try again in {N} days {N} hours {N} minutes.」 — Codex URL: https://didreset.com/zh/errors/codex-you-have-hit-your-usage-limit-try-again-in-days-hours-minutes Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade to Pro (https://openai.com/chatgpt/pricing) or try again in {N} days {N} hours {N} minutes. 「You've hit your usage limit…try again in {N} days {N} hours {N} minutes.」是 Codex 限额提示的 2025 年版本，用相对倒计时。以天计的倒计时说明用完的是周限额；它按你账号的时钟刷新，不是本站板面追踪的全员重置。最近一次 Codex 官方重置：{lastResetDate}。 「You've hit your usage limit. Upgrade to Pro (https://openai.com/chatgpt/pricing) or try again in {N} days {N} hours {N} minutes.」是什么意思？ 这句话来自 2025 年 9 月的 Codex CLI 0.39.0（openai/codex issue #3899，2025-09-19）：和现在同一个限额拦截，只是等待时间写成「N days N hours N minutes」，而不是绝对的「try again at」时间。到 2026 年文案已改成绝对时间戳（issue #34360 显示「try again at Jul 25, 2026, 11:09 AM」），但搜索里仍大量出现旧句子。 报告里的倒计时是「5 days 22 hours 11 minutes」。几天的等待不可能是 5 小时窗口，所以这种形式的提示说的是周限额已用完；OpenAI 按「a five-hour window and a weekly window」计量 Codex，且没有公布周限额数字。 2025 年文案只给了升级 Pro 一条付费路线；新版本还会提到 credits，以及有活动时的重置卡或购买重置。 满的是哪个窗口？ - 窗口: 5 小时窗口 · 何时刷新: 从打开窗口的第一条消息起 5 小时，按账号计算。OpenAI 原话：「You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining.」2026-07-12 对 Plus、Business、Pro 暂时取消；Plus 自 2026-08-25 恢复（原帖写「Tomorrow」，帖子时间 2026-08-25 01:16 UTC，即太平洋时间 8 月 25 日）；Pro $100 / $200「for the upcoming months」不启用。 · 在哪里看: 在 Codex CLI 会话里输入 /status（「5h limit」行）；App 的 Settings → Usage；chatgpt.com/codex/settings/usage - 窗口: 周限额 · 何时刷新: 每周一次，按你账号自己的日期。OpenAI 没有公布周限额数字（原话「Weekly limits may also apply」）。兑换整张重置卡会「changes your weekly reset date」；购买即时重置后，下一次自动周刷新是你第一条请求之后的第 7 天。 · 在哪里看: /status（「Weekly limit」行）；Settings → Usage；chatgpt.com/codex/settings/usage 看到它该怎么办 1. 把倒计时当作周限额的等待: 倒计时里有「days」就说明用完的是周限额。用当前时间加上倒计时得到刷新时刻，或者升级 CLI，让 /status 直接在「Weekly limit」旁打印绝对时间。 2. 在新版 CLI 里 /status: 在会话里输入 /status；OpenAI 帮助中心把它作为「Check which allowance is exhausted, your credit balance, and any reset time shown.」的方式。App 的 Settings → Usage 和 chatgpt.com/codex/settings/usage 显示同样的进度条。 3. 升级、加 credits、兑换重置，或者等: 2025 年的文案只提供 Pro，但 OpenAI 现在的说法是「You may be able to add credits, use an available reset, upgrade, or wait until the displayed reset time.」等待永远可选；OpenAI Support 不会重置额度。 什么时候重置？ 你自己的窗口：倒计时就是你账号的周刷新，从文案出现那一刻起算。把算出的时间贴进窗口倒计时得到提醒和日历文件，或者在新版 CLI 里 /status 读精确时间。 全员重置：OpenAI 偶尔会重置所有人的 Codex 额度，「is applied directly to the usage limits covered by the announcement」。这类事件就是上方板面行和 Codex 历史页记录的内容；它们没有排期，所以几天的倒计时只会被这样的公告、重置卡或购买重置缩短。 FAQ: Q: 为什么显示的是天数而不是时间？ A: 因为这是 2025 年 9 月的文案版本，把等待写成「N days N hours N minutes」（issue #3899）。2026 年的 Codex CLI 改为显示绝对时间，例如「try again at Jul 25, 2026, 11:09 AM」（issue #34360）。 Q: 五天的倒计时是哪个窗口？ A: 周限额。Codex 同时计 5 小时窗口和周窗口，「5 days 22 hours 11 minutes」的等待不可能是 5 小时那个。OpenAI 没有公布周限额数字，只说「Weekly limits may also apply.」 Q: 这条文案还在用吗？ A: 这个措辞来自 CLI 0.39.0（2025-09-19）。之后的报告都用绝对的「try again at」形式，所以看到倒计时版本说明你的 CLI 较旧；背后的限额是一样的。 Q: 全员重置能缩短倒计时吗？ A: 能，前提是 OpenAI 宣布了：自动重置「is applied directly to the usage limits covered by the announcement」。上方板面行显示有记录的最近一次；它不能预测下一次。 Sources: 1. openai/codex issue #3899 (2025-09-19): Codex CLI 0.39.0 (Windows, gpt-5-codex): "try again in 5 days 22 hours 11 minutes" — https://github.com/openai/codex/issues/3899 (verified 2026-09-17) 2. openai/codex issue #34360 (2026-07-20): the 2026 wording of the same notice uses an absolute time: "try again at Jul 25, 2026, 11:09 AM" — https://github.com/openai/codex/issues/34360 (verified 2026-09-17) 3. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 4. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 5. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 6. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 7. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 8. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 9. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### "5h limit: [████████████████░░░░] 82% left (resets 22:41) Weekly limit: [███████████████████░] 97% left (resets 17:41 on 1 Jul)" — Codex URL: https://didreset.com/errors/codex-status-5h-limit-weekly-limit-resets Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: 5h limit: [████████████████░░░░] 82% left (resets 22:41) Weekly limit: [███████████████████░] 97% left (resets 17:41 on 1 Jul) The "5h limit" and "Weekly limit" lines are the two usage bars Codex CLI prints when you type /status inside a session: the first shows how much of your 5-hour window is left and the clock time it refreshes, the second your weekly limit and its refresh date and time. Both run on your account's own clock; they are not the account-wide reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "5h limit: [████████████████░░░░] 82% left (resets 22:41) Weekly limit: [███████████████████░] 97% left (resets 17:41 on 1 Jul)" mean? Typing /status in an active Codex CLI session prints a box with the model, directory, permissions, account and session id, then two bars: "5h limit" with a percentage left and a "resets HH:MM" clock time, and "Weekly limit" with a percentage left and a "resets HH:MM on D Mon" date. OpenAI's help center names this as the way to check: "In an active Codex CLI session, enter /status." The two lines are the two windows OpenAI meters, "a five-hour window and a weekly window"; you need allowance in both to keep working, and you can hit the five-hour bar before five hours have passed. The box header adds "Visit https://chatgpt.com/codex/settings/usage for up-to-date information on rate limits and credits". Some accounts show a third "premium limit" row (issue #30349). The source box (openai/codex issue #29948, CLI 0.142.0) is also a reminder that the bars and the limit notice can disagree: the same session was blocked with "You've hit your usage limit" while showing 82% and 97% left, which OpenAI labelled bug / rate-limits. Which window is full? - Window: 5-hour window · Resets: Five hours after the first message that opened the window, per account. OpenAI: "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." Removed for Plus, Business and Pro on 2026-07-12; back for Plus since 2026-08-25 (the post says "Tomorrow" and is stamped 2026-08-25 01:16 UTC, i.e. Pacific time 2026-08-25); "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. · Where to check: /status inside an active Codex CLI session (the "5h limit" line); Settings → Usage in the app; chatgpt.com/codex/settings/usage - Window: Weekly limit · Resets: Once a week on your account's own date. OpenAI publishes no weekly figure ("Weekly limits may also apply"). A full banked reset "changes your weekly reset date"; after a purchased reset the next automatic weekly reset is seven days after your first request. · Where to check: /status (the "Weekly limit" line); Settings → Usage; chatgpt.com/codex/settings/usage What to do when you see it 1. Type /status inside the session: Start codex, then type /status at the prompt (not "codex /status" in the shell). The docs describe it as "Display session configuration and token usage." 2. Read the two reset values: The time after "resets" on the 5h line is when your 5-hour window refreshes; the "resets HH:MM on D Mon" on the Weekly line is your weekly refresh. OpenAI does not say which time zone the clock uses, so compare it with chatgpt.com/codex/settings/usage if in doubt. 3. Put the times on a countdown: Paste each reset time into the window tracker to get a countdown and a calendar file for your own windows. /usage, the other command, lets you "View account token usage or use a rate-limit reset" if you hold a banked one. 4. Report bars with room that still block you: If both bars show allowance and Codex still says you have hit your limit, that is the bug in issues #29948 and #30349: file an issue at github.com/openai/codex with the /status box, or contact OpenAI Support with the reset time shown and your time zone. When does it reset? Your own windows: exactly what the two lines say. The 5-hour bar refills at the "resets" time on its line; the weekly bar refills at the date and time on its line, and a full banked reset or a purchased reset moves that date. Paste the times into the window tracker for a countdown. Account-wide resets: when OpenAI announces a reset for everyone it "is applied directly to the usage limits covered by the announcement", so both bars jump without you doing anything. Those announcements are what the board row above and the Codex history page record; they are not scheduled. FAQ: Q: Is the time after "resets" in my local time zone? A: OpenAI's documentation does not say. The box in issue #29948 shows plain clock times ("resets 22:41", "resets 17:41 on 1 Jul") with no zone. Compare them with the reset time on chatgpt.com/codex/settings/usage, which the box itself points to. Q: What is the difference between /status and /usage? A: Per the Codex CLI docs, "/status — Display session configuration and token usage" prints the box with the two limit bars, while "/usage — View account token usage or use a rate-limit reset" opens a menu to inspect token activity or redeem an available banked reset. Q: Why was I blocked with 82% left? A: The box in issue #29948 was captured right before "You've hit your usage limit. Upgrade to Plus …" appeared on a Business account, and #30349 shows the same at 99%/100% on Plus; OpenAI labelled both bug / rate-limits. When the bars have room and you are still blocked, report it rather than waiting. Q: Does a global Codex reset change these bars? A: Yes, immediately: an announced reset "is applied directly to the usage limits covered by the announcement" and "won't appear as a saved reset". Check the board row above for the last reset on record. Sources: 1. openai/codex issue #29948 (2026-06-25): /status box from Codex CLI 0.142.0 showing both limit bars, then "You've hit your usage limit" with 82% and 97% left — https://github.com/openai/codex/issues/29948 (verified 2026-09-17) 2. openai/codex issue #30349 (2026-06-27): /status box with a third "premium limit" row on a Plus account — https://github.com/openai/codex/issues/30349 (verified 2026-09-17) 3. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 4. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 5. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 6. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 7. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 9. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### 「5h limit: [████████████████░░░░] 82% left (resets 22:41) Weekly limit: [███████████████████░] 97% left (resets 17:41 on 1 Jul)」 — Codex URL: https://didreset.com/zh/errors/codex-status-5h-limit-weekly-limit-resets Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: 5h limit: [████████████████░░░░] 82% left (resets 22:41) Weekly limit: [███████████████████░] 97% left (resets 17:41 on 1 Jul) 「5h limit」和「Weekly limit」是你在 Codex CLI 会话里输入 /status 后打印的两条用量条：第一条是 5 小时窗口剩余比例和刷新时刻，第二条是周限额剩余比例和刷新日期时间。两条都按你账号自己的时钟走，不是本站出发板追踪的全员重置。有记录的最近一次 Codex 官方重置是 {lastResetDate}。 「5h limit: [████████████████░░░░] 82% left (resets 22:41) Weekly limit: [███████████████████░] 97% left (resets 17:41 on 1 Jul)」是什么意思？ 在 Codex CLI 会话里输入 /status，会打印一个框：模型、目录、权限、账号、会话 id，然后是两条进度条：「5h limit」带剩余百分比和「resets HH:MM」时刻，「Weekly limit」带剩余百分比和「resets HH:MM on D Mon」日期。OpenAI 帮助中心把它列为查限额的方式：「In an active Codex CLI session, enter /status.」 这两行对应 OpenAI 计量的两个窗口「a five-hour window and a weekly window」：两边都要有余量才能继续，而且不到 5 小时也可能先撞到 5 小时那条。框顶还有一句「Visit https://chatgpt.com/codex/settings/usage for up-to-date information on rate limits and credits」。部分账号会多一行「premium limit」（issue #30349）。 出处的那个框（openai/codex issue #29948，CLI 0.142.0）也提醒你进度条和拦截提示可能不一致：同一个会话在显示 82% 和 97% 剩余时被「You've hit your usage limit」拦住，OpenAI 打了 bug / rate-limits 标签。 满的是哪个窗口？ - 窗口: 5 小时窗口 · 何时刷新: 从打开窗口的第一条消息起 5 小时，按账号计算。OpenAI 原话：「You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining.」2026-07-12 对 Plus、Business、Pro 暂时取消；Plus 自 2026-08-25 恢复（原帖写「Tomorrow」，帖子时间 2026-08-25 01:16 UTC，即太平洋时间 8 月 25 日）；Pro $100 / $200「for the upcoming months」不启用。 · 在哪里看: 在 Codex CLI 会话里输入 /status（「5h limit」行）；App 的 Settings → Usage；chatgpt.com/codex/settings/usage - 窗口: 周限额 · 何时刷新: 每周一次，按你账号自己的日期。OpenAI 没有公布周限额数字（原话「Weekly limits may also apply」）。兑换整张重置卡会「changes your weekly reset date」；购买即时重置后，下一次自动周刷新是你第一条请求之后的第 7 天。 · 在哪里看: /status（「Weekly limit」行）；Settings → Usage；chatgpt.com/codex/settings/usage 看到它该怎么办 1. 在会话里输入 /status: 先启动 codex，然后在提示符里输入 /status（不是在 shell 里跑「codex /status」）。文档对它的描述是「Display session configuration and token usage.」 2. 读两处 resets 的值: 5h 行「resets」后面的时刻是 5 小时窗口的刷新时间；Weekly 行的「resets HH:MM on D Mon」是周刷新。OpenAI 没有说明这个时钟用哪个时区，拿不准就和 chatgpt.com/codex/settings/usage 上的时间对一下。 3. 把时间放进倒计时: 把两个刷新时间分别贴进窗口倒计时，得到你自己窗口的倒计时和日历文件。另一个命令 /usage 可以「View account token usage or use a rate-limit reset」，手里有重置卡时用它兑换。 4. 有余量却被拦就报告: 两条进度条都有余量、Codex 却说你到了上限，就是 issue #29948 和 #30349 里的 bug：带 /status 框去 github.com/openai/codex 提 issue，或联系 OpenAI Support，附上显示的重置时间和你的时区。 什么时候重置？ 你自己的窗口：就是这两行写的时间。5 小时条在它那行的「resets」时刻回满；周限额条在它那行的日期时间回满，兑换整张重置卡或购买重置会把这个日期提前。把时间贴进窗口倒计时即可。 全员重置：OpenAI 宣布全员重置时，「is applied directly to the usage limits covered by the announcement」，两条进度条会自己跳上去，不需要你操作。这类公告就是上方板面行和 Codex 历史页记录的内容；它们没有排期。 FAQ: Q: 「resets」后面的时间是我的本地时区吗？ A: OpenAI 的文档没有说明。issue #29948 里的框只显示纯时刻（「resets 22:41」「resets 17:41 on 1 Jul」），没有时区。和框里指向的 chatgpt.com/codex/settings/usage 上的重置时间对照一下。 Q: /status 和 /usage 有什么区别？ A: 按 Codex CLI 文档：「/status — Display session configuration and token usage」打印带两条限额条的框；「/usage — View account token usage or use a rate-limit reset」打开一个菜单，可以查看 token 活动或兑换可用的重置卡。 Q: 为什么剩 82% 就被拦了？ A: issue #29948 的框是在 Business 账号出现「You've hit your usage limit. Upgrade to Plus …」之前截的，#30349 在 Plus 账号上 99%/100% 时也一样；OpenAI 都打了 bug / rate-limits 标签。进度条有余量却被拦时，去报告，不要干等。 Q: 全员重置会改变这两条进度条吗？ A: 会，而且是立即：宣布的重置「is applied directly to the usage limits covered by the announcement」，并且「won't appear as a saved reset」。看上方板面行有记录的最近一次重置。 Sources: 1. openai/codex issue #29948 (2026-06-25): /status box from Codex CLI 0.142.0 showing both limit bars, then "You've hit your usage limit" with 82% and 97% left — https://github.com/openai/codex/issues/29948 (verified 2026-09-17) 2. openai/codex issue #30349 (2026-06-27): /status box with a third "premium limit" row on a Plus account — https://github.com/openai/codex/issues/30349 (verified 2026-09-17) 3. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 4. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 5. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 6. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 7. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 9. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5-hour limit back for Plus "Tomorrow", stays off for Pro $100 and $200 — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### "You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days." — Codex URL: https://didreset.com/errors/codex-you-have-a-new-rate-limit-reset-available Language: en · Surface: App · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days. "You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days." is the Codex app banner that appears when OpenAI grants you a banked reset: a one-time reset saved to your account that you redeem yourself. It refreshes your own 5-hour and weekly windows when you use it; it is not the account-wide reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days." mean? The banner announces a banked reset. OpenAI defines it as "a one-time Codex usage-limit reset saved to your account until you use it or it expires. It is not purchased credit and does not permanently increase your plan's usage limit." Banked resets are promotional: launch rewards or successful referrals, not something you can normally buy. Nothing happens until you redeem it. A banked reset "must be applied by you"; when you do, "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." It "is consumed only when it successfully refreshes at least one eligible usage window", and it expires: the Codex pricing page says banked resets "are usable for 30 days after they're granted". In the source report (openai/codex issue #37396, 2026-08-07, Codex app 26.803.41515 on macOS) a Pro user saw this banner while the usage panel kept showing 1 available reset instead of 2, and the count went from 1 to 0 after redeeming, so the granted reset never landed. OpenAI does not replace resets that were "not applied as described" without investigating, so report a missing one. Which window is full? - Window: 5-hour window · Resets: Refreshed the moment you redeem a full banked reset: "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows". · Where to check: Settings → Usage after redeeming; /status in the CLI - Window: Weekly limit · Resets: Refreshed by the same redemption, and the reset "changes your weekly reset date"; the new date is shown in Settings → Usage. · Where to check: Settings → Usage; /status (the "Weekly limit" line) What to do when you see it 1. Confirm the count: Open the profile menu and look at the usage summary, or open Settings → Usage in the desktop app, web or CLI. OpenAI: "select the usage summary showing the available reset count, such as 1 reset available." If the count did not change after the banner, that is the issue #37396 bug. 2. Redeem it when it will refresh something: Pick "1 reset available" or "Full reset", "Review the expiration details, then confirm that you want to use it." A full reset refreshes both windows and moves your weekly reset date; if there is nothing to reset, it stays available. In the CLI, /usage lets you "redeem an available earned reset". 3. Watch the expiry: Banked resets are usable for 30 days after they are granted, and "Once an unused reset expires, it cannot be restored or reissued." Support does not offer replacements for resets you let expire. 4. Report a reset that never arrived: If the banner appeared but the count is unchanged, contact OpenAI Support so it can review eligibility, or file an issue at github.com/openai/codex with screenshots of the banner and the usage panel. "Reporting an issue alone does not qualify you for an additional reset or compensation." When does it reset? Your own windows: this banner does not reset anything by itself. When you redeem the banked reset, your 5-hour and weekly windows refresh immediately and Settings → Usage shows the new weekly reset date; the dashboard "may take a short time to update". Until then the windows keep their normal clocks, which /status shows. Account-wide resets: a banked reset is different from the announced resets on this board. OpenAI: an automatic or global reset "does not create a saved reset"; it is applied to everyone at once. The board row above shows the last such reset on record; the Codex history page lists every announcement, including the banked-reset drops of 2026-09-03 and 2026-09-04. FAQ: Q: Do I have to use the reset right away? A: No, but within 30 days: "Banked rate-limit resets are usable for 30 days after they're granted." OpenAI also says expiration can vary by offer, so check the date shown in your account. Tibo Sottiaux noted some users "had stacked up to three banked resets". Q: What exactly does a full banked reset refresh? A: "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." It is consumed only if at least one window is actually refreshed; otherwise it remains available. Q: The banner appeared but my available-reset count did not change. Why? A: That is the bug in issue #37396: a Pro user saw the banner, the panel stayed at 1 available, and redeeming took it to 0. Contact OpenAI Support with screenshots so eligibility can be reviewed; the help center says Support "can review eligibility" but does not reissue resets on request alone. Q: Can I redeem it from the CLI or on mobile? A: Yes. In Codex CLI, "/usage — View account token usage or use a rate-limit reset"; on 2026-07-12 Tibo Sottiaux announced "the ability to use a banked reset from web and mobile", where before it was desktop-app only. Sources: 1. openai/codex issue #37396 (2026-08-07): Pro user on Codex app 26.803.41515 (macOS) saw the banner but the available-reset count stayed at 1 instead of rising to 2 — https://github.com/openai/codex/issues/37396 (verified 2026-09-17) 2. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 3. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 4. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 5. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 6. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 7. Tibo Sottiaux on X, 2026-07-12: banked reset can now be used from web and mobile — https://x.com/thsottiaux/status/2076418567143408112 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-06-28: some users had stacked up to three banked resets — https://x.com/thsottiaux/status/2071381664853319742 (verified 2026-09-17) 9. OpenAI Developer Community: OpenAI_Support reply on how banked resets work (2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) ### 「You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days.」 — Codex URL: https://didreset.com/zh/errors/codex-you-have-a-new-rate-limit-reset-available Language: zh · Surface: App · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days. 「You have a new rate limit reset available…」是 Codex App 在 OpenAI 发给你一张重置卡时显示的横幅：一次性重置，存在账号里，由你自己兑换。兑换时刷新的是你自己的 5 小时窗口和周窗口；不是本站板面追踪的全员重置。有记录的最近一次 Codex 官方重置：{lastResetDate}。 「You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days.」是什么意思？ 这条横幅宣布的是一张重置卡。OpenAI 的定义：「a one-time Codex usage-limit reset saved to your account until you use it or it expires. It is not purchased credit and does not permanently increase your plan's usage limit.」重置卡是促销性质的：发布奖励或推荐成功，通常买不到。 不兑换什么都不会发生。重置卡「must be applied by you」；兑换时「Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date.」它「is consumed only when it successfully refreshes at least one eligible usage window」，而且会过期：Codex 定价页写明重置卡「are usable for 30 days after they're granted」。 出处 openai/codex issue #37396（2026-08-07，macOS 上的 Codex App 26.803.41515）里，Pro 用户看到横幅后用量面板仍显示 1 张可用而不是 2 张，兑换后从 1 变 0，也就是说新发的那张根本没入账。OpenAI 不会不经调查就补发「not applied as described」的重置，所以缺了就去报告。 满的是哪个窗口？ - 窗口: 5 小时窗口 · 何时刷新: 兑换整张重置卡的那一刻刷新：「Using a full banked reset refreshes your 5-hour and weekly Codex usage windows」。 · 在哪里看: 兑换后看 Settings → Usage；CLI 里 /status - 窗口: 周限额 · 何时刷新: 同一次兑换一起刷新，并且「changes your weekly reset date」；新日期显示在 Settings → Usage。 · 在哪里看: Settings → Usage；/status 的「Weekly limit」行 看到它该怎么办 1. 先确认数量: 打开个人菜单看用量摘要，或在桌面 App、网页、CLI 打开 Settings → Usage。OpenAI 原话：「select the usage summary showing the available reset count, such as 1 reset available.」横幅出现后数量没变，就是 issue #37396 的 bug。 2. 在它能刷新到东西的时候兑换: 点「1 reset available」或「Full reset」，「Review the expiration details, then confirm that you want to use it.」整张重置刷新两个窗口并改变周刷新日期；没有可刷新的内容时它会继续保留。CLI 里用 /usage 可以「redeem an available earned reset」。 3. 盯住过期时间: 重置卡自发放起 30 天内可用，「Once an unused reset expires, it cannot be restored or reissued.」让它过期的，Support 不会补发。 4. 没入账就报告: 横幅出现但数量没变，联系 OpenAI Support 让其核查资格，或带横幅和用量面板截图去 github.com/openai/codex 提 issue。「Reporting an issue alone does not qualify you for an additional reset or compensation.」 什么时候重置？ 你自己的窗口：这条横幅本身不重置任何东西。兑换重置卡时，5 小时窗口和周窗口立即刷新，Settings → Usage 显示新的周刷新日期；面板「may take a short time to update」。在那之前两个窗口按原来的时钟走，/status 能看到。 全员重置：重置卡和本站板面上的公告重置不是一回事。OpenAI：自动或全员重置「does not create a saved reset」，它一次性作用于所有人。上方板面行显示有记录的最近一次；Codex 历史页列出每一次公告，包括 2026-09-03 和 2026-09-04 的重置卡发放。 FAQ: Q: 必须马上用掉吗？ A: 不必，但要在 30 天内：「Banked rate-limit resets are usable for 30 days after they're granted.」OpenAI 也说过期时间可能因活动而异，以账号里显示的日期为准。Tibo Sottiaux 提到有用户「had stacked up to three banked resets」。 Q: 整张重置卡到底刷新什么？ A: 「Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date.」只有至少一个窗口真的被刷新，它才会被消耗；否则继续可用。 Q: 横幅出现了，但可用数量没变，怎么回事？ A: 这就是 issue #37396 的 bug：Pro 用户看到横幅，面板仍是 1 张，兑换后变 0。带截图联系 OpenAI Support 核查资格；帮助中心说 Support「can review eligibility」，但不会仅凭申请就补发。 Q: 能在 CLI 或手机上兑换吗？ A: 能。Codex CLI 里「/usage — View account token usage or use a rate-limit reset」；Tibo Sottiaux 在 2026-07-12 宣布「the ability to use a banked reset from web and mobile」，此前只能在桌面 App 里兑换。 Sources: 1. openai/codex issue #37396 (2026-08-07): Pro user on Codex app 26.803.41515 (macOS) saw the banner but the available-reset count stayed at 1 instead of rising to 2 — https://github.com/openai/codex/issues/37396 (verified 2026-09-17) 2. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 3. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 4. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 5. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 6. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 7. Tibo Sottiaux on X, 2026-07-12: banked reset can now be used from web and mobile — https://x.com/thsottiaux/status/2076418567143408112 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-06-28: some users had stacked up to three banked resets — https://x.com/thsottiaux/status/2071381664853319742 (verified 2026-09-17) 9. OpenAI Developer Community: OpenAI_Support reply on how banked resets work (2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) ### "1 reset available" — Codex URL: https://didreset.com/errors/codex-1-reset-available-full-reset-banked-reset Language: en · Surface: App · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: 1 reset available Variants: Full reset "1 reset available" in the Codex usage summary means a banked reset is stored on your account: "a one-time Codex usage-limit reset saved to your account until you use it or it expires." Redeeming a full reset refreshes your own 5-hour and weekly windows and moves your weekly reset date; it is not the account-wide reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "1 reset available" mean? The label sits in the profile menu's usage summary in the Codex desktop app, on the web and in the CLI's /usage menu. OpenAI's help center: "Select the usage summary showing an available reset, such as '1 reset available' or 'Full reset.'" The number is how many banked resets you hold. A banked reset is not credit and does not raise your plan's limit. Redeeming it does one thing: "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." It "is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available." They come from promotions. OpenAI Support: "Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase." The help center records drops to Plus, Pro and Business users on 2026-09-03 and 2026-09-04 around the GPT-6 Astra launch, alongside a separate global reset that was applied immediately and could not be banked. Banked resets are usable for 30 days after they are granted. Which window is full? - Window: 5-hour window · Resets: Refreshed the moment you redeem a full banked reset: "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows". · Where to check: Settings → Usage after redeeming; /status in the CLI - Window: Weekly limit · Resets: Refreshed by the same redemption, and the reset "changes your weekly reset date"; the new date is shown in Settings → Usage. · Where to check: Settings → Usage; /status (the "Weekly limit" line) What to do when you see it 1. Open the usage summary: "Open settings → usage on the desktop app, CLI, or on web", or select the usage summary in the profile menu that shows "1 reset available" or "Full reset". In Codex CLI, "Type /usage to open the usage menu. Choose whether to show token activity or redeem an available earned reset." 2. Check the expiry, then confirm: "Review the expiration details, then confirm that you want to use it." Banked resets are usable for 30 days after they are granted; "Once an unused reset expires, it cannot be restored or reissued." 3. Verify the new reset time: "Check Settings → Usage for your updated reset time." The dashboard "may take a short time to update. If the percentage does not change immediately, refresh or reopen Settings → Usage". 4. Report a reset that was spent without refreshing: If the count dropped but your bars did not move, contact OpenAI Support with screenshots; a reset "is consumed only when it successfully refreshes at least one eligible usage window", so a spent reset with no effect is a bug worth reporting (compare issue #37396). When does it reset? Your own windows: nothing changes until you redeem. Once you do, both windows refresh at once and Settings → Usage shows the new weekly reset date, which /status also prints; a purchased instant reset behaves the same way but cannot be saved. Put the new time into the window tracker for a countdown. Account-wide resets: an announced global reset "does not create a saved reset"; it is applied to everyone immediately and shows up on this board as a reset event. The board row above shows the last one on record; the Codex history page lists every announcement, including banked-reset drops, with their original posts. FAQ: Q: What happens if I redeem it when nothing needs resetting? A: Nothing is lost: "The reset is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available." Redeem it when a window is actually low. Q: Does my weekly reset date move? A: Yes. A full banked reset "changes your weekly reset date", and the help center tells you to "Check Settings → Usage for your updated reset time." The 5-hour window is refreshed at the same time. Q: Where do banked resets come from, and can I buy one? A: From promotions: "Eligible Plus and Pro users may receive them through limited promotions, such as launch rewards or successful referrals." They are not normally purchasable; what OpenAI does sell to eligible personal Plus and Pro accounts is a purchased instant reset, which "cannot be saved for later". Q: Can I redeem it from the CLI, web or mobile? A: CLI: /usage lets you "redeem an available earned reset". Web and mobile: since 2026-07-12, when Tibo Sottiaux announced "the ability to use a banked reset from web and mobile. Before today you could only do it in the desktop app". Sources: 1. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 3. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 4. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 5. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 6. Tibo Sottiaux on X, 2026-07-12: banked reset can now be used from web and mobile — https://x.com/thsottiaux/status/2076418567143408112 (verified 2026-09-17) 7. OpenAI Developer Community: OpenAI_Support reply on how banked resets work (2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) 8. openai/codex issue #37396 (2026-08-07): Pro user whose available-reset count stayed at 1 after a new banked reset was announced — https://github.com/openai/codex/issues/37396 (verified 2026-09-17) 9. openai/codex issue #27934 (2026-06-12): request to let Linux users redeem banked resets from the web or CLI — https://github.com/openai/codex/issues/27934 (verified 2026-09-17) ### 「1 reset available」 — Codex URL: https://didreset.com/zh/errors/codex-1-reset-available-full-reset-banked-reset Language: zh · Surface: App · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: 1 reset available Variants: Full reset 「1 reset available」出现在 Codex 用量摘要里，表示账号里存着一张重置卡：一次性的 Codex 额度重置，存到你用掉或过期为止。兑换整张重置卡会刷新你的 5 小时窗口和周窗口并改变周刷新日期；不是本站板面追踪的全员重置。有记录的最近一次 Codex 官方重置：{lastResetDate}。 「1 reset available」是什么意思？ 这个标签出现在 Codex 桌面 App、网页的个人菜单用量摘要里，以及 CLI 的 /usage 菜单。OpenAI 帮助中心原话：「Select the usage summary showing an available reset, such as '1 reset available' or 'Full reset.'」数字就是你持有的重置卡张数。 重置卡不是 credits，也不会提高计划上限。兑换它只做一件事：「Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date.」它「is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available.」 重置卡来自促销活动。OpenAI Support：「Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase.」帮助中心记载了 GPT-6 Astra 发布前后 2026-09-03 与 2026-09-04 两次向 Plus、Pro、Business 用户的发放，同时还有一次立即生效、不可存储的全员重置。重置卡自发放起 30 天内可用。 满的是哪个窗口？ - 窗口: 5 小时窗口 · 何时刷新: 兑换整张重置卡的那一刻刷新：「Using a full banked reset refreshes your 5-hour and weekly Codex usage windows」。 · 在哪里看: 兑换后看 Settings → Usage；CLI 里 /status - 窗口: 周限额 · 何时刷新: 同一次兑换一起刷新，并且「changes your weekly reset date」；新日期显示在 Settings → Usage。 · 在哪里看: Settings → Usage；/status 的「Weekly limit」行 看到它该怎么办 1. 打开用量摘要: 「Open settings → usage on the desktop app, CLI, or on web」，或在个人菜单里点显示「1 reset available」或「Full reset」的用量摘要。Codex CLI 里：「Type /usage to open the usage menu. Choose whether to show token activity or redeem an available earned reset.」 2. 看过期时间，再确认: 「Review the expiration details, then confirm that you want to use it.」重置卡自发放起 30 天内可用；「Once an unused reset expires, it cannot be restored or reissued.」 3. 核对新的刷新时间: 「Check Settings → Usage for your updated reset time.」面板「may take a short time to update. If the percentage does not change immediately, refresh or reopen Settings → Usage」。 4. 被消耗却没刷新就报告: 数量减了但进度条没动，带截图联系 OpenAI Support；重置卡「is consumed only when it successfully refreshes at least one eligible usage window」，花掉却无效属于 bug，值得报告（参考 issue #37396）。 什么时候重置？ 你自己的窗口：不兑换什么都不变。兑换后两个窗口同时刷新，Settings → Usage 显示新的周刷新日期，/status 也会打印；购买的即时重置效果相同但不能存。把新时间贴进窗口倒计时即可。 全员重置：宣布的全员重置「does not create a saved reset」，它立即作用于所有人，并以重置事件的形式出现在本站板面上。上方板面行显示有记录的最近一次；Codex 历史页列出每一次公告（含重置卡发放）和原帖。 FAQ: Q: 没有需要重置的东西时兑换会怎样？ A: 不会白白损失：「The reset is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available.」等某个窗口真的不多了再用。 Q: 我的周刷新日期会变吗？ A: 会。整张重置卡「changes your weekly reset date」，帮助中心让你「Check Settings → Usage for your updated reset time.」5 小时窗口同时刷新。 Q: 重置卡从哪来，能买吗？ A: 来自促销：「Eligible Plus and Pro users may receive them through limited promotions, such as launch rewards or successful referrals.」通常买不到；OpenAI 向符合条件的个人 Plus、Pro 账号出售的是购买即时重置，它「cannot be saved for later」。 Q: 能在 CLI、网页或手机上兑换吗？ A: CLI：/usage 可以「redeem an available earned reset」。网页和手机：自 2026-07-12 起，Tibo Sottiaux 宣布「the ability to use a banked reset from web and mobile. Before today you could only do it in the desktop app」。 Sources: 1. OpenAI Help Center: How banked Codex resets work (article 20001498) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) 3. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (article 20001516) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 4. ChatGPT Learn: Codex pricing and usage limits — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 5. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 6. Tibo Sottiaux on X, 2026-07-12: banked reset can now be used from web and mobile — https://x.com/thsottiaux/status/2076418567143408112 (verified 2026-09-17) 7. OpenAI Developer Community: OpenAI_Support reply on how banked resets work (2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) 8. openai/codex issue #37396 (2026-08-07): Pro user whose available-reset count stayed at 1 after a new banked reset was announced — https://github.com/openai/codex/issues/37396 (verified 2026-09-17) 9. openai/codex issue #27934 (2026-06-12): request to let Linux users redeem banked resets from the web or CLI — https://github.com/openai/codex/issues/27934 (verified 2026-09-17) ### "unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses" — Codex URL: https://didreset.com/errors/codex-unexpected-status-401-unauthorized Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses Variants: 401 Unauthorized: Missing bearer authentication | Your access token could not be refreshed because your refresh token was revoked. Please log out and sign in again. | Missing scopes: api.responses.write "unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses" is not a usage-limit message. HTTP 401 means Codex sent a request without a valid credential: OpenAI's error reference lists 401 as "Invalid Authentication". No 5-hour window, weekly limit, banked reset or account-wide reset changes it; the fix is codex logout and codex login. The last official Codex reset on record was {lastResetDate}, and it is irrelevant here. What does "unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses" mean? A 401 is an authentication error, not a quota. OpenAI's API error reference lists "401 - Invalid Authentication" with the solution "Ensure the correct API key and requesting organization are being used"; rate limiting is a different code, "429 - Rate limit reached for requests". When Codex prints this line, the request to chatgpt.com/backend-api/codex/responses was rejected before any usage was counted. It is about your cached login. Codex "caches login details locally in a plaintext file at ~/.codex/auth.json or in your OS-specific credential store" and "refreshes tokens automatically during use before they expire". The reports behind this page show what breaks that: a revoked refresh token (issue #41973: "Your access token could not be refreshed because your refresh token was revoked. Please log out and sign in again."), a stale token missing a scope (issue #33664: "Missing scopes: api.responses.write", fixed by renaming auth.json and signing in again), or a desktop app stuck without a bearer token (issue #30039: "401 Unauthorized: Missing bearer authentication"). Because nothing is being metered, the usage-limit tools do not apply: /status may even show full bars, a banked reset has nothing to refresh, and an account-wide Codex reset does not touch credentials. Treat "You've hit your usage limit" and "401 Unauthorized" as two unrelated messages. Which window is full? - Window: None: 401 is authentication, not a usage window · Resets: It does not reset. Waiting for the 5-hour window or weekly limit, redeeming a banked reset, or an account-wide Codex reset will not clear a 401. Only a valid credential does. · Where to check: codex login status ("Logged in using ChatGPT" or the API-key state); the Account line in /status What to do when you see it 1. Check what Codex thinks you are signed in as: Run codex login status; the docs say it shows "the active authentication method" and exits 0 when credentials are present. Note that in issue #41973 it still reported "Logged in using ChatGPT" while every request failed with 401, so a green status does not rule out a revoked token. 2. Sign out, then sign in again: Run codex logout ("Remove saved credentials for both API key and ChatGPT authentication") or type /logout in the session, then codex login and complete the browser flow; codex login --device-auth is the alternative when the browser hand-off fails. In the desktop app or IDE, use Log out in the profile menu and sign in again. 3. If it persists, clear the cached credential file: Codex stores credentials in ~/.codex/auth.json (or the OS credential store, per cli_auth_credentials_store). In issue #33664 the 401 survived reinstalling the app and only went away after renaming auth.json and signing in fresh. Also confirm no OPENAI_API_KEY variable or ~/.codex/.env is overriding your ChatGPT login, since API-key sign-in is a separate path. 4. Do not wait for a reset: Nothing about this error is time-based. If a fresh login still returns 401, file an issue at github.com/openai/codex with the request id and cf-ray printed after the message (label auth), or contact OpenAI Support. When does it reset? Your own windows: not involved. A 401 is returned before Codex counts anything against your 5-hour window or weekly limit, so /status can show room while every request fails. Re-authenticating is the only fix; the window tracker and the reset time in a limit notice have no bearing on it. Account-wide resets: also not involved. An announced Codex reset "is applied directly to the usage limits covered by the announcement"; it does not reissue tokens or scopes. The board row above is shown for context only. If you see both a 401 and "You've hit your usage limit", fix the login first, then read the limit page. FAQ: Q: Is 401 Unauthorized a rate limit? A: No. In OpenAI's error reference 401 is "Invalid Authentication" and rate limiting is 429 ("Rate limit reached for requests"). A 401 means the request carried no valid credential; no usage was consumed and no window is involved. Q: Will waiting for my window or a Codex reset fix it? A: No. Usage windows and account-wide resets change allowances, not credentials. The reports behind this page were resolved by signing out and back in or by replacing the cached auth.json, never by waiting. Q: codex login status says I am logged in. Why do I still get 401? A: A cached credential can be present but no longer accepted. Issue #41973 shows "Logged in using ChatGPT" alongside the 401, with the app also reporting that the refresh token was revoked; issue #33664 had a token missing the api.responses.write scope. Sign out and sign in again, or rename ~/.codex/auth.json and log in fresh. Q: How is this different from "You've hit your usage limit"? A: That message is a usage limit with a reset time and options to add credits, use a reset, upgrade or wait. A 401 has none of those: it is rejected before metering. OpenAI's help center tells limited users to check "any reset time shown"; a 401 shows no reset time because there is nothing to wait for. Sources: 1. openai/codex issue #26497 (2026-06-05): Codex CLI 0.137.0: "unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses" — https://github.com/openai/codex/issues/26497 (verified 2026-09-18) 2. openai/codex issue #41973 (2026-09-01): Codex CLI 0.152.0 and Desktop on macOS: same 401 after a successful ChatGPT login; "Your access token could not be refreshed because your refresh token was revoked. Please log out and sign in again." — https://github.com/openai/codex/issues/41973 (verified 2026-09-18) 3. openai/codex issue #30039 (2026-06-25): Codex Desktop 26.616.81150: "401 Unauthorized: Missing bearer authentication", then 429 after retries, no sign-in prompt — https://github.com/openai/codex/issues/30039 (verified 2026-09-18) 4. openai/codex issue #33664 (2026-07-16): Codex Desktop on Windows: "401 Unauthorized / Missing scopes: api.responses.write"; fixed by renaming ~/.codex/auth.json and signing in again — https://github.com/openai/codex/issues/33664 (verified 2026-09-18) 5. ChatGPT Learn: Authentication (codex login, codex login status, codex logout, login caching) — https://learn.chatgpt.com/docs/auth (verified 2026-09-18) 6. OpenAI Developers: API error codes (401 Invalid Authentication, 429 Rate limit reached) — https://developers.openai.com/api/docs/guides/error-codes (verified 2026-09-18) 7. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 8. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) ### 「unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses」 — Codex URL: https://didreset.com/zh/errors/codex-unexpected-status-401-unauthorized Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses Variants: 401 Unauthorized: Missing bearer authentication | Your access token could not be refreshed because your refresh token was revoked. Please log out and sign in again. | Missing scopes: api.responses.write 「unexpected status 401 Unauthorized…」不是额度提示：Codex 的请求没有带有效凭据，OpenAI 错误码页称为「Invalid Authentication」。等窗口、重置卡、全员重置都没用，解决办法是退出后重新 codex login。最近一次 Codex 官方重置：{lastResetDate}。 「unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses」是什么意思？ 401 是认证错误，不是额度。OpenAI 的 API 错误码页把「401 - Invalid Authentication」的解决办法写为「Ensure the correct API key and requesting organization are being used」；限流是另一个码，「429 - Rate limit reached for requests」。Codex 打印这一行时，发往 chatgpt.com/backend-api/codex/responses 的请求在计入任何用量之前就被拒绝了。 问题出在你缓存的登录信息。Codex「caches login details locally in a plaintext file at ~/.codex/auth.json or in your OS-specific credential store」，并且「refreshes tokens automatically during use before they expire」。本页背后的报告展示了它会怎么坏：refresh token 被吊销（issue #41973：「Your access token could not be refreshed because your refresh token was revoked. Please log out and sign in again.」）、缓存的 token 缺少 scope（issue #33664：「Missing scopes: api.responses.write」，重命名 auth.json 重新登录后解决）、桌面 App 卡在没有 bearer token 的状态（issue #30039：「401 Unauthorized: Missing bearer authentication」）。 因为根本没有计量，限额相关的工具都不适用：/status 甚至可能显示进度条满格，重置卡没有东西可刷新，全员 Codex 重置也不碰凭据。把「You've hit your usage limit」和「401 Unauthorized」当作两条互不相干的信息。 满的是哪个窗口？ - 窗口: 无：401 是认证，不是用量窗口 · 何时刷新: 它不会「重置」。等 5 小时窗口或周限额、兑换重置卡、等全员 Codex 重置都清不掉 401，只有有效凭据才行。 · 在哪里看: codex login status（「Logged in using ChatGPT」或 API key 状态）；/status 里的 Account 行 看到它该怎么办 1. 先看 Codex 认为你是谁: 运行 codex login status；文档说它显示「the active authentication method」，有凭据时退出码为 0。注意 issue #41973 里它仍报告「Logged in using ChatGPT」，但每个请求都 401，所以状态正常不代表 token 没被吊销。 2. 退出登录，再登录: 运行 codex logout（「Remove saved credentials for both API key and ChatGPT authentication」）或在会话里输入 /logout，再 codex login 完成浏览器授权；浏览器回跳失败时用 codex login --device-auth。桌面 App 或 IDE 里在个人菜单点 Log out 后重新登录。 3. 还不行就清掉缓存的凭据文件: Codex 把凭据存在 ~/.codex/auth.json（或按 cli_auth_credentials_store 存在系统凭据库）。issue #33664 里重装 App 都没用，重命名 auth.json 重新登录后 401 才消失。同时确认没有 OPENAI_API_KEY 环境变量或 ~/.codex/.env 覆盖了你的 ChatGPT 登录，因为 API key 登录是另一条路径。 4. 不要等重置: 这个错误和时间无关。重新登录后仍然 401，就带文案后面打印的 request id 和 cf-ray 去 github.com/openai/codex 提 issue（标签 auth），或联系 OpenAI Support。 什么时候重置？ 你自己的窗口：不相关。401 在 Codex 把任何东西计入 5 小时窗口或周限额之前就返回了，所以 /status 可以显示有余量而每个请求都失败。重新认证是唯一的办法；窗口倒计时和限额提示里的重置时间对它没有意义。 全员重置：同样不相关。宣布的 Codex 重置「is applied directly to the usage limits covered by the announcement」，不会重新签发 token 或 scope。上方板面行只是顺带展示。如果你同时看到 401 和「You've hit your usage limit」，先修登录，再看限额页。 FAQ: Q: 401 Unauthorized 是限流吗？ A: 不是。OpenAI 错误码页里 401 是「Invalid Authentication」，限流是 429（「Rate limit reached for requests」）。401 表示请求没有带有效凭据；没有消耗任何用量，也不涉及任何窗口。 Q: 等窗口刷新或等 Codex 重置能修好吗？ A: 不能。用量窗口和全员重置改变的是额度，不是凭据。本页背后的报告都是靠退出再登录、或替换缓存的 auth.json 解决的，没有一例是等好的。 Q: codex login status 显示已登录，为什么还是 401？ A: 缓存的凭据可以存在但不再被接受。issue #41973 里「Logged in using ChatGPT」和 401 同时出现，App 还提示 refresh token 已被吊销；issue #33664 的 token 缺少 api.responses.write scope。退出再登录，或者重命名 ~/.codex/auth.json 后重新登录。 Q: 它和「You've hit your usage limit」有什么区别？ A: 那条是限额提示，带重置时间和加 credits、用重置、升级或等待的选项。401 一样都没有：它在计量之前就被拒绝。OpenAI 帮助中心让被限额的用户看「any reset time shown」；401 不显示重置时间，因为没有什么可等。 Sources: 1. openai/codex issue #26497 (2026-06-05): Codex CLI 0.137.0: "unexpected status 401 Unauthorized: {"detail":"Unauthorized"}, url: https://chatgpt.com/backend-api/codex/responses" — https://github.com/openai/codex/issues/26497 (verified 2026-09-18) 2. openai/codex issue #41973 (2026-09-01): Codex CLI 0.152.0 and Desktop on macOS: same 401 after a successful ChatGPT login; "Your access token could not be refreshed because your refresh token was revoked. Please log out and sign in again." — https://github.com/openai/codex/issues/41973 (verified 2026-09-18) 3. openai/codex issue #30039 (2026-06-25): Codex Desktop 26.616.81150: "401 Unauthorized: Missing bearer authentication", then 429 after retries, no sign-in prompt — https://github.com/openai/codex/issues/30039 (verified 2026-09-18) 4. openai/codex issue #33664 (2026-07-16): Codex Desktop on Windows: "401 Unauthorized / Missing scopes: api.responses.write"; fixed by renaming ~/.codex/auth.json and signing in again — https://github.com/openai/codex/issues/33664 (verified 2026-09-18) 5. ChatGPT Learn: Authentication (codex login, codex login status, codex logout, login caching) — https://learn.chatgpt.com/docs/auth (verified 2026-09-18) 6. OpenAI Developers: API error codes (401 Invalid Authentication, 429 Rate limit reached) — https://developers.openai.com/api/docs/guides/error-codes (verified 2026-09-18) 7. ChatGPT Learn: Codex CLI slash commands (/status, /usage, /logout) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-18) 8. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) ### "You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after {H:MM AM/PM}" — ChatGPT URL: https://didreset.com/errors/chatgpt-you-have-reached-the-current-usage-cap-for-gpt-4 Language: en · Surface: Web · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after {H:MM AM/PM} "You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after {H:MM AM/PM}" was ChatGPT's banner for the GPT-4 message cap on Plus in 2023–2024. The time in the banner is when your own cap lifted; the banner offered the default model meanwhile. It is a ChatGPT chat limit, not the Codex reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after {H:MM AM/PM}" mean? This is a historical ChatGPT banner. It appeared on the web app when a Plus account used up its GPT-4 message allowance; the banner let you keep chatting with the default model and gave the clock time after which GPT-4 would answer again. The community topic that preserves the wording was opened on 2023-11-20 and ran to 89 posts. Users in that thread describe the cap as 40 messages per 3 hours and complain of hitting it after fewer than 10 messages; in April 2024 free-tier users reported seeing the same banner. Those figures are user reports, not an OpenAI statement, and OpenAI's current help pages do not carry this sentence. It is not about Codex. OpenAI's Codex help article keeps ChatGPT-side limits apart from Codex usage: "ChatGPT file uploads, image generation, or voice have separate usage limits, reset periods, and banners" and "those limits do not apply to Codex". The GPT-4 cap was a ChatGPT chat limit; the board on this page tracks account-wide Codex resets, not chat caps. Which window is full? - Window: GPT-4 message cap (ChatGPT Plus, 2023–2024) · Resets: At the time after "try again after" in the banner. Users in the source thread reported 40 messages per 3 hours; OpenAI has not published a current figure for this banner. · Where to check: The banner itself; there is no /status command in ChatGPT chat What to do when you see it 1. Note the time in the banner: The "try again after" clock time is when your GPT-4 cap lifted. Waiting was the only way to get GPT-4 back on the same account; OpenAI Support does not reset ChatGPT usage limits. 2. Continue with the default model: The banner's own suggestion: keep the conversation going on the default model and switch back when the time passes. 3. Upgrade if you hit it constantly: OpenAI's help center points to chatgpt.com/pricing for "higher-usage options for individuals"; the exact GPT-4 cap per plan was not published on the pages verified for this site. When does it reset? Your own cap: the banner gave the clock time; that was your account's own cap window, not a schedule shared with anyone else. If you still see a time-based ChatGPT banner today, the same rule applies: the time shown is when your own limit lifts. Account-wide resets: the board row above records official Codex resets announced by OpenAI, which refresh Codex usage windows for everyone. They have no connection to the GPT-4 chat cap, and this site has no record of an announced reset for ChatGPT chat caps. FAQ: Q: Is this message still current? A: It comes from the 2023–2024 ChatGPT Plus interface; the thread preserving it was opened on 2023-11-20. Newer ChatGPT limit banners use different wording, and OpenAI's current Codex help article does not contain this sentence. Treat it as historical. Q: How many GPT-4 messages did the cap allow? A: Users in the source thread cite 40 messages per 3 hours and report being cut off after fewer than 10. That number is a community report; OpenAI did not publish a figure on the pages verified for this site, so this page does not state one as fact. Q: Does this banner affect Codex? A: No. It was a ChatGPT chat cap. OpenAI's Codex help article says ChatGPT-side limits such as "file uploads, image generation, or voice have separate usage limits, reset periods, and banners" and "do not apply to Codex". Codex has its own 5-hour and weekly windows, checked with /status. Q: Can OpenAI Support reset the cap? A: No: "OpenAI Support does not reset ChatGPT or Codex usage limits. If you reach a limit, wait until it resets or use another available option shown in your account." Sources: 1. OpenAI Developer Community topic 514923 (2023-11-20): "You've reached the current usage cap for GPT-4 …" (89 posts; users cite 40 messages per 3 hours) — https://community.openai.com/t/youve-reached-the-current-usage-cap-for-gpt-4-you-can-continue-with-the-default-model-now-or-try-again-after-xx-xx-am-pm/514923 (verified 2026-09-18) 2. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) ### 「You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after {H:MM AM/PM}」 — ChatGPT URL: https://didreset.com/zh/errors/chatgpt-you-have-reached-the-current-usage-cap-for-gpt-4 Language: zh · Surface: 网页 · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after {H:MM AM/PM} 「You've reached the current usage cap for GPT-4…」是 2023–2024 年 ChatGPT Plus 网页版触到 GPT-4 消息上限时的横幅。横幅里的时间是你自己的上限解除时刻，期间可换默认模型。它是 ChatGPT 聊天上限，不是本站追踪的 Codex 重置。最近一次：{lastResetDate}。 「You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after {H:MM AM/PM}」是什么意思？ 这是一条历史横幅。Plus 账号在网页版用完 GPT-4 消息额度时出现，横幅允许你用默认模型继续聊，并给出 GPT-4 恢复回答的时刻。保留这句原文的社区主题开于 2023-11-20，累计 89 帖。 主题里用户把上限描述为每 3 小时 40 条，并抱怨不到 10 条就被拦；2024 年 4 月免费用户也报告看到同样的横幅。这些数字是用户说法，不是 OpenAI 的声明，OpenAI 现行帮助页面里也没有这句话。 它与 Codex 无关。OpenAI 的 Codex 帮助文章把 ChatGPT 侧的限制与 Codex 用量分开：「ChatGPT file uploads, image generation, or voice have separate usage limits, reset periods, and banners」，并且「those limits do not apply to Codex」。GPT-4 上限是 ChatGPT 的聊天上限；本页板面追踪的是 Codex 的全员重置，不是聊天上限。 满的是哪个窗口？ - 窗口: GPT-4 消息上限（ChatGPT Plus，2023–2024） · 何时刷新: 横幅里「try again after」之后的时刻。来源主题里用户报告为每 3 小时 40 条；OpenAI 没有公布这条横幅的现行数字。 · 在哪里看: 横幅本身；ChatGPT 聊天里没有 /status 命令 看到它该怎么办 1. 记下横幅里的时间: 「try again after」的时刻就是你的 GPT-4 上限解除的时间。同一账号要拿回 GPT-4 只能等；OpenAI Support 不会重置 ChatGPT 额度。 2. 用默认模型继续: 这是横幅自己给的建议：用默认模型把对话继续下去，时间过了再切回来。 3. 经常撞上就升级: OpenAI 帮助中心把「higher-usage options for individuals」指向 chatgpt.com/pricing；各计划的 GPT-4 具体上限没有在本站核实过的页面上公布。 什么时候重置？ 你自己的上限：横幅给出了时刻，那是你账号自己的上限窗口，不是和别人共享的排期。今天如果还看到带时间的 ChatGPT 横幅，规则一样：显示的时间就是你自己的限制解除时间。 全员重置：上方板面行记录的是 OpenAI 宣布的官方 Codex 重置，它为所有人刷新 Codex 用量窗口。它和 GPT-4 聊天上限没有关系，本站也没有 ChatGPT 聊天上限被宣布重置的记录。 FAQ: Q: 这条文案现在还在用吗？ A: 它来自 2023–2024 年的 ChatGPT Plus 界面；保留它的主题开于 2023-11-20。之后的 ChatGPT 限额横幅措辞不同，OpenAI 现行的 Codex 帮助文章里也没有这句话。当作历史文案看待。 Q: GPT-4 上限允许多少条消息？ A: 来源主题里用户说每 3 小时 40 条，并报告不到 10 条就被拦。这是社区说法；OpenAI 没有在本站核实过的页面上公布数字，所以本页不把它当事实陈述。 Q: 这条横幅影响 Codex 吗？ A: 不影响。它是 ChatGPT 的聊天上限。OpenAI 的 Codex 帮助文章说 ChatGPT 侧的限制，如「file uploads, image generation, or voice have separate usage limits, reset periods, and banners」，并且「do not apply to Codex」。Codex 有自己的 5 小时窗口和周窗口，用 /status 查。 Q: OpenAI Support 能重置这个上限吗？ A: 不能：「OpenAI Support does not reset ChatGPT or Codex usage limits. If you reach a limit, wait until it resets or use another available option shown in your account.」 Sources: 1. OpenAI Developer Community topic 514923 (2023-11-20): "You've reached the current usage cap for GPT-4 …" (89 posts; users cite 40 messages per 3 hours) — https://community.openai.com/t/youve-reached-the-current-usage-cap-for-gpt-4-you-can-continue-with-the-default-model-now-or-try-again-after-xx-xx-am-pm/514923 (verified 2026-09-18) 2. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) ### "You've hit the free plan limit for image generation and editing. You'll be able to make this change once the limit resets in {N} hours and {N} minutes." — ChatGPT URL: https://didreset.com/errors/chatgpt-you-have-hit-the-free-plan-limit-for-image-generation Language: en · Surface: Web · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit the free plan limit for image generation and editing. You'll be able to make this change once the limit resets in {N} hours and {N} minutes. Variants: You've hit the free plan limit for image generations and edits. You'll be able to update this image in {N} hours and {N} minutes once the limit resets. "You've hit the free plan limit for image generation and editing…" is ChatGPT's banner when a Free account has used its image allowance. The countdown in the banner is that limit's own clock; OpenAI says image limits "have separate usage limits, reset periods, and banners" and "do not apply to Codex", so it is not the Codex reset this board tracks. The last official Codex reset on record was {lastResetDate}. What does "You've hit the free plan limit for image generation and editing. You'll be able to make this change once the limit resets in {N} hours and {N} minutes." mean? This banner belongs to ChatGPT's image feature on the Free plan. It appears when you ask for an image generation or edit after the plan's image allowance is used up, and it carries its own countdown to the next reset; the message itself is the only place the time is shown. It is separate from text messages and from Codex. OpenAI's help center: "ChatGPT file uploads, image generation, or voice have separate usage limits, reset periods, and banners", giving examples such as "50 images in the last day" and "resets in 720 hours", and states that "those limits do not apply to Codex". The source report (community topic 1292047, 2025-06-18) shows the countdown can stick: the poster saw "15 hours and 27 minutes" and, a full day later, the same 15 hours and 27 minutes in a slightly different wording. OpenAI has not published the size of the Free image allowance on the pages verified for this site. Which window is full? - Window: Image generation and editing (ChatGPT Free) · Resets: At the end of the countdown in the banner; it has its own reset period, separate from text messages. Allowance size: not published on the pages verified for this site. · Where to check: The banner itself; OpenAI's image & video generation limits article is linked from the Codex help article What to do when you see it 1. Let the countdown run: The "resets in {N} hours and {N} minutes" figure is the image limit's own clock. No text-message or Codex window is involved, and OpenAI Support does not reset ChatGPT usage limits. 2. Upgrade for a larger image allowance: The banner names the free plan; paid ChatGPT plans have different image limits. OpenAI points to chatgpt.com/pricing for plan options and to its image & video generation limits article for details. 3. Contact Support if the countdown never moves: If the same number reappears day after day, as in the source thread, OpenAI's guidance applies: contact Support when "access has not returned after the stated reset time", with a screenshot and the time it happened including your time zone. When does it reset? Your own limit: when the countdown in the banner reaches zero. It is a per-account image allowance with "separate usage limits, reset periods, and banners", so nothing you do in Codex or in text chat changes it. Account-wide resets: the board row above records official Codex resets announced by OpenAI, which refresh Codex usage windows for everyone; image limits are explicitly outside Codex ("those limits do not apply to Codex"), and this site has no record of an announced reset for ChatGPT image limits. FAQ: Q: Why does the countdown show the same time a day later? A: That is what the source report describes: "15 hours and 27 minutes" on 2025-06-18 and again on 2025-06-19. OpenAI has not explained it. Its help center says to contact Support if access has not returned after the stated reset time. Q: Does this limit block Codex or text messages? A: No. "ChatGPT file uploads, image generation, or voice have separate usage limits, reset periods, and banners", and "those limits do not apply to Codex". Text chat and Codex keep working while the image limit counts down. Q: How many free images do I get? A: Not published on the pages verified for this site. OpenAI's Codex help article only gives examples of image banners ("50 images in the last day", "resets in 720 hours") and links to a separate image & video generation limits article for current figures. Q: Can OpenAI Support reset it? A: No: "OpenAI Support does not reset ChatGPT or Codex usage limits." Support investigates when usage looks miscounted or access has not returned after the stated reset time. Sources: 1. OpenAI Developer Community topic 1292047 (2025-06-18): "You've hit the free plan limit for image generation and editing …" with the same 15 hours and 27 minutes countdown a day later — https://community.openai.com/t/youve-hit-the-free-plan-limit/1292047 (verified 2026-09-18) 2. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) ### 「You've hit the free plan limit for image generation and editing. You'll be able to make this change once the limit resets in {N} hours and {N} minutes.」 — ChatGPT URL: https://didreset.com/zh/errors/chatgpt-you-have-hit-the-free-plan-limit-for-image-generation Language: zh · Surface: 网页 · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit the free plan limit for image generation and editing. You'll be able to make this change once the limit resets in {N} hours and {N} minutes. Variants: You've hit the free plan limit for image generations and edits. You'll be able to update this image in {N} hours and {N} minutes once the limit resets. 「You've hit the free plan limit for image generation and editing…」是 ChatGPT 免费账号用完图片额度时的横幅。倒计时是该限额自己的时钟；OpenAI 说图片限额「do not apply to Codex」，它不是本站追踪的 Codex 重置。最近一次：{lastResetDate}。 「You've hit the free plan limit for image generation and editing. You'll be able to make this change once the limit resets in {N} hours and {N} minutes.」是什么意思？ 这条横幅属于 ChatGPT 免费计划的图片功能。当图片额度用完后再请求生成或编辑图片时出现，自带到下一次刷新的倒计时；这个时间只显示在横幅里。 它与文字消息和 Codex 都是分开的。OpenAI 帮助中心：「ChatGPT file uploads, image generation, or voice have separate usage limits, reset periods, and banners」，并举了「50 images in the last day」「resets in 720 hours」这样的例子，明确「those limits do not apply to Codex」。 出处（社区主题 1292047，2025-06-18）显示倒计时可能卡住：发帖人看到「15 hours and 27 minutes」，整整一天后仍是 15 小时 27 分钟，只是措辞略有不同。OpenAI 没有在本站核实过的页面上公布免费图片额度的大小。 满的是哪个窗口？ - 窗口: 图片生成与编辑（ChatGPT Free） · 何时刷新: 横幅里的倒计时结束时；它有自己的刷新周期，与文字消息分开。额度大小：本站核实过的页面上未公布。 · 在哪里看: 横幅本身；Codex 帮助文章里链接了 OpenAI 的图片与视频生成限额文章 看到它该怎么办 1. 让倒计时走完: 「resets in {N} hours and {N} minutes」是图片限额自己的时钟。不涉及文字消息或 Codex 的窗口，OpenAI Support 也不会重置 ChatGPT 额度。 2. 要更多图片额度就升级: 横幅点名了免费计划；付费 ChatGPT 计划的图片限额不同。OpenAI 把计划选项指向 chatgpt.com/pricing，细节见其图片与视频生成限额文章。 3. 倒计时一直不动就联系 Support: 像来源主题那样同一个数字天天重现，适用 OpenAI 的指引：在「access has not returned after the stated reset time」时联系 Support，附上截图和发生时间（含时区）。 什么时候重置？ 你自己的限额：横幅里的倒计时归零时。这是按账号计的图片额度，「separate usage limits, reset periods, and banners」，你在 Codex 或文字聊天里做什么都不会改变它。 全员重置：上方板面行记录的是 OpenAI 宣布的官方 Codex 重置，它为所有人刷新 Codex 用量窗口；图片限额明确不在 Codex 之内（「those limits do not apply to Codex」），本站也没有 ChatGPT 图片限额被宣布重置的记录。 FAQ: Q: 为什么一天后倒计时还是同一个数字？ A: 来源报告就是这样：2025-06-18 显示「15 hours and 27 minutes」，2025-06-19 还是这个数。OpenAI 没有解释。帮助中心说过了显示的重置时间仍无法使用时联系 Support。 Q: 这个限额会挡住 Codex 或文字消息吗？ A: 不会。「ChatGPT file uploads, image generation, or voice have separate usage limits, reset periods, and banners」，并且「those limits do not apply to Codex」。图片倒计时期间文字聊天和 Codex 照常可用。 Q: 免费能生成多少张图？ A: 本站核实过的页面上未公布。OpenAI 的 Codex 帮助文章只举了图片横幅的例子（「50 images in the last day」「resets in 720 hours」），并链接到单独的图片与视频生成限额文章查看现行数字。 Q: OpenAI Support 能重置它吗？ A: 不能：「OpenAI Support does not reset ChatGPT or Codex usage limits.」只有用量疑似算错，或过了显示的重置时间仍无法使用时，Support 才会调查。 Sources: 1. OpenAI Developer Community topic 1292047 (2025-06-18): "You've hit the free plan limit for image generation and editing …" with the same 15 hours and 27 minutes countdown a day later — https://community.openai.com/t/youve-hit-the-free-plan-limit/1292047 (verified 2026-09-18) 2. OpenAI Help Center: Using Codex with your ChatGPT plan (article 11369540) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-18) ### "You've hit your usage limit. Upgrade for more usage. Get more credits on higher plans." — Cursor URL: https://didreset.com/errors/cursor-you-have-hit-your-usage-limit-upgrade-for-more-usage Language: en · Surface: IDE · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade for more usage. Get more credits on higher plans. "You've hit your usage limit. Upgrade for more usage. Get more credits on higher plans." is Cursor's notice when your plan's included monthly usage is used up. Cursor: "Usage resets monthly with your billing cycle. Unused usage does not roll over. The reset date is shown on the Spending tab." It is Cursor's own meter, not the account-wide Codex reset this board tracks; the last official Codex reset on record was {lastResetDate}. What does "You've hit your usage limit. Upgrade for more usage. Get more credits on higher plans." mean? This message comes from the Agent panel in the Cursor IDE. Cursor plans include monthly usage pools ("Cursor Models" and "Other Models"); when the included usage is gone, Cursor says "You'll see a notification in the editor. You can either enable on-demand usage (pay-as-you-go) or upgrade to a higher plan." The source report (Cursor forum topic 144570, 2025-11-28, Cursor 2.1.39 on a $20 Pro plan) saw the message while running gpt-5-pro with only about 70% of the month's usage consumed and one day before the billing period ended; enabling on-demand usage let the request through. A Cursor ambassador (community group CursorAmbassadors, not Cursor staff) replied on 2025-12-01 that the plan had renewed on November 29 and "you should have access to all models now since the usage limit has reset!" Cursor's meter is monthly and per billing cycle. It has no 5-hour or weekly window, and it is not connected to OpenAI's Codex resets: the board row on this page is shown because this site tracks Codex, not because a Codex reset changes Cursor's included usage. Which window is full? - Window: Included monthly usage (Cursor Models pool + Other Models pool) · Resets: "Usage resets monthly with your billing cycle. Unused usage does not roll over." · Where to check: The Spending tab in the Cursor dashboard: "It shows real-time usage for both pools, remaining allowance, and any on-demand charges." "The reset date is shown on the Spending tab." - Window: Team plans · Resets: "For teams, all members' usage resets at the same time based on the team billing cycle." · Where to check: The Spending tab What to do when you see it 1. Open the Spending tab: In the Cursor dashboard, "Go to the Spending tab in your dashboard. It shows real-time usage for both pools, remaining allowance, and any on-demand charges." The reset date is shown there too. 2. Enable on-demand usage or upgrade: Cursor's two documented ways forward: "Enable on-demand usage: Pay for additional requests at the same API rates" or "Upgrade your plan: Higher-tier plans include more usage". Cursor Models (Cursor Grok 4.6, Cursor Grok 4.5, Composer 2.5) draw from a separate pool that may still have room. 3. Or wait for the billing cycle: Included usage comes back on your billing date; the source report was blocked one day before renewal and had access again after it. 4. Ask on the forum with a request ID if the numbers look wrong: In the source thread the ambassador asked for the request ID before checking with the team; the question of why the block came at 70% usage was left unanswered. When does it reset? Your own allowance: on your billing date, shown on the Spending tab. "Usage resets monthly with your billing cycle. Unused usage does not roll over." There is no shorter window to wait for; on-demand usage or an upgrade is the only way to continue before then. Account-wide resets: the board row above is OpenAI's Codex, the product this site tracks. An official Codex reset refreshes Codex usage windows; it is a different meter from Cursor's monthly pools, and this site has no record of Cursor announcing account-wide resets. FAQ: Q: Why did it fire at 70% of my monthly usage? A: The source thread does not resolve that. The reporter was running gpt-5-pro one day before the billing period ended; on-demand usage let the request through, and the ambassador only confirmed that the limit had reset after the November 29 renewal. Q: Does unused usage roll over to next month? A: No. Cursor: "Usage resets monthly with your billing cycle. Unused usage does not roll over. The reset date is shown on the Spending tab." What you do not use in a billing period is simply gone when the next one starts. Q: Does the Codex board on this page have anything to do with Cursor's limit? A: No. The board tracks OpenAI's account-wide Codex resets; Cursor meters its own monthly pools per billing cycle. The board is shown here because this site follows Codex, Claude Code and Grok, not because a Codex reset refreshes Cursor usage. Q: Do requests with my own API key count against the limit? A: On individual plans, no: "Your provider bills you directly for the model cost, and those requests don't draw from either pool." On Teams and Enterprise plans they still carry the Cursor Token Rate of $0.25 per million tokens under Other Models. Sources: 1. Cursor Forum topic 144570 (2025-11-28): "You've hit your usage limit" on Cursor 2.1.39 Pro with 70% used; a Cursor ambassador confirmed the reset at the 2025-11-29 renewal — https://forum.cursor.com/t/youve-hit-your-usage-limit-when-only-70-used/144570 (verified 2026-09-18) 2. Cursor Docs: Usage and limits (monthly reset with billing cycle, Spending tab) — https://cursor.com/help/models-and-usage/usage-limits (verified 2026-09-18) ### 「You've hit your usage limit. Upgrade for more usage. Get more credits on higher plans.」 — Cursor URL: https://didreset.com/zh/errors/cursor-you-have-hit-your-usage-limit-upgrade-for-more-usage Language: zh · Surface: IDE · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your usage limit. Upgrade for more usage. Get more credits on higher plans. 「You've hit your usage limit. Upgrade for more usage…」是 Cursor 在计划当月包含用量用完时的提示。Cursor 文档：用量随账单周期按月刷新，不结转，刷新日期在 Spending 页。这是 Cursor 自己的计量，不是本站追踪的 Codex 重置；最近一次：{lastResetDate}。 「You've hit your usage limit. Upgrade for more usage. Get more credits on higher plans.」是什么意思？ 这条提示出现在 Cursor IDE 的 Agent 面板。Cursor 计划按月包含用量池（「Cursor Models」和「Other Models」）；包含用量用完时，Cursor 说「You'll see a notification in the editor. You can either enable on-demand usage (pay-as-you-go) or upgrade to a higher plan.」 出处（Cursor 论坛主题 144570，2025-11-28，$20 Pro 计划上的 Cursor 2.1.39）是在跑 gpt-5-pro 时看到这句话，当月用量只用了约 70%，离账单周期结束还有一天；打开 on-demand usage 后请求就能通过。一位 Cursor 大使（社区组 CursorAmbassadors，不是 Cursor 员工）在 2025-12-01 回复计划已于 11 月 29 日续费，「you should have access to all models now since the usage limit has reset!」 Cursor 的计量是按月、按账单周期的。它没有 5 小时或周窗口，也和 OpenAI 的 Codex 重置没有联系：本页显示 Codex 板面行是因为本站追踪 Codex，不是因为 Codex 重置会改变 Cursor 的包含用量。 满的是哪个窗口？ - 窗口: 当月包含用量（Cursor Models 池 + Other Models 池） · 何时刷新: 「Usage resets monthly with your billing cycle. Unused usage does not roll over.」 · 在哪里看: Cursor 控制台的 Spending 页：「It shows real-time usage for both pools, remaining allowance, and any on-demand charges.」「The reset date is shown on the Spending tab.」 - 窗口: 团队计划 · 何时刷新: 「For teams, all members' usage resets at the same time based on the team billing cycle.」 · 在哪里看: Spending 页 看到它该怎么办 1. 打开 Spending 页: 在 Cursor 控制台里，「Go to the Spending tab in your dashboard. It shows real-time usage for both pools, remaining allowance, and any on-demand charges.」刷新日期也在那里。 2. 打开 on-demand usage 或升级: Cursor 文档给的两条路：「Enable on-demand usage: Pay for additional requests at the same API rates」或「Upgrade your plan: Higher-tier plans include more usage」。Cursor Models（Cursor Grok 4.6、Cursor Grok 4.5、Composer 2.5）走单独的池子，可能还有余量。 3. 或者等账单周期: 包含用量在你的账单日回来；出处报告在续费前一天被拦，续费后恢复。 4. 数字不对就带 request ID 去论坛问: 出处主题里大使先要了 request ID 再找团队核查；至于为什么 70% 就被拦，主题没有给出答案。 什么时候重置？ 你自己的额度：在你的账单日，Spending 页有显示。「Usage resets monthly with your billing cycle. Unused usage does not roll over.」没有更短的窗口可等；在那之前只能靠 on-demand usage 或升级继续。 全员重置：上方板面行是 OpenAI 的 Codex，本站追踪的产品。Codex 官方重置刷新的是 Codex 用量窗口；它和 Cursor 的月度用量池是两套计量，本站也没有 Cursor 宣布全员重置的记录。 FAQ: Q: 为什么用到 70% 就被拦了？ A: 出处主题没有解决这个问题。报告者在账单周期结束前一天跑 gpt-5-pro；打开 on-demand usage 后请求通过，大使只确认了 11 月 29 日续费后限额已重置。 Q: 没用完的用量会结转到下个月吗？ A: 不会。Cursor：「Usage resets monthly with your billing cycle. Unused usage does not roll over. The reset date is shown on the Spending tab.」 Q: 本页的 Codex 板面和 Cursor 的限额有关系吗？ A: 没有。板面追踪的是 OpenAI 面向全员的 Codex 重置；Cursor 按账单周期计量自己的月度用量池。这里显示板面是因为本站追踪 Codex、Claude Code 和 Grok，不是因为 Codex 重置会刷新 Cursor 用量。 Q: 用自己的 API key 发的请求算进限额吗？ A: 个人计划不算：「Your provider bills you directly for the model cost, and those requests don't draw from either pool.」Teams 和 Enterprise 计划仍按 Other Models 收取每百万 token $0.25 的 Cursor Token Rate。 Sources: 1. Cursor Forum topic 144570 (2025-11-28): "You've hit your usage limit" on Cursor 2.1.39 Pro with 70% used; a Cursor ambassador confirmed the reset at the 2025-11-29 renewal — https://forum.cursor.com/t/youve-hit-your-usage-limit-when-only-70-used/144570 (verified 2026-09-18) 2. Cursor Docs: Usage and limits (monthly reset with billing cycle, Spending tab) — https://cursor.com/help/models-and-usage/usage-limits (verified 2026-09-18) ### "You've hit your session limit · resets {time}" — Claude Code URL: https://didreset.com/errors/claude-code-you-have-hit-your-session-limit-resets Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your session limit · resets {time} Variants: You've hit your session limit · resets 3:45pm | You've hit your session limit · resets 1pm (America/Los_Angeles) "You've hit your session limit · resets {time}" is Claude Code's message when your rolling session allowance is used up. Anthropic says the session limit resets every five hours, and the time in the message is your own account's clock: run /usage to see it. Since v2.1.234 Claude Code can wait and continue automatically. It is not the account-wide reset this board tracks; the board row below shows the last one on record. What does "You've hit your session limit · resets {time}" mean? Subscription plans include a rolling usage allowance. When the session part of it runs out, Claude Code shows this message and blocks further requests until the reset time it names. The official Error reference lists it next to three siblings: the weekly, Opus and Sonnet limits. The session and weekly limits are shared across all models, so switching models with /model does not restore access. The docs also warn that usage counts against the session and weekly allowances at the same time, so one burst of heavy activity, such as a large workflow fanout, can exhaust the weekly allowance before the session window resets. Since v2.1.234, an interactive session signed in with a claude.ai subscription can wait in place: a line at the bottom reads "Usage limit reached · continuing automatically at 3:45pm · esc to cancel", and after the reset the task picks up where it stopped. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai What to do when you see it 1. Read the time in the message: That is your own reset time, not a site-wide event. Wait for it, or leave the session open and let Claude Code wait for you. 2. Turn on automatic continue: In the CLI the setting is "Continue automatically at usage limit" in /config. In the Desktop app's Code tab, the session-limit card has an "Auto-continue when limits reset" checkbox. The two are separate settings. 3. Run /usage: It shows your plan limits and when each one resets, so you can tell a session limit from a weekly one. 4. Buy extra usage if you cannot wait: On Pro and Max, /usage-credits turns on usage credits. On Team and Enterprise it requests them from your admin. Credits do not move the reset time. 5. Upgrade for higher base limits: Anthropic lists plan sizes at claude.com/pricing; Max 5x and Max 20x include five and twenty times the Pro per-session allowance. When does it reset? Your own window: Anthropic states that the session-based usage limit resets every five hours, and the exact time is the one in the message and in /usage. Usage credits do not change it: "Usage credits don't affect this reset timing." Put that time on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those announcements are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site never promises one; subscribe to alerts to hear about the next one. FAQ: Q: Does switching models fix "You've hit your session limit"? A: No. The Error reference says the session and weekly limits are shared across all models, so switching models does not restore access. Only the model-family messages ("You've hit your Opus limit", "You've hit your Sonnet limit") can be worked around by switching to a model outside that family with /model. Q: Can Claude Code continue by itself after the reset? A: Yes, since v2.1.234, in an interactive session signed in with a claude.ai subscription. While it waits, the bottom line reads "Usage limit reached · continuing automatically at 3:45pm · esc to cancel"; press Esc at an empty prompt to cancel. You can turn the behavior off with "Continue automatically at usage limit" in /config. Q: Is the session limit the same as the weekly limit? A: No. The session limit resets every five hours. The weekly limit resets at a fixed time each week that is assigned to your account. Usage counts against both at once, so a heavy burst can exhaust the weekly allowance before the session window resets; /usage shows both bars. Q: Is this the same as an official Claude reset on the board? A: No. The message is about your own five-hour window. The board tracks account-wide resets that Anthropic announces publicly, each linked to its source post. If the row above says a reset is on record, that is a separate event from the time in your message. Sources: 1. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 2. Claude Code docs: Interactive mode — Wait for a usage limit to reset — https://code.claude.com/docs/en/interactive-mode#wait-for-a-usage-limit-to-reset (verified 2026-09-18) 3. anthropics/claude-code issue #93148 (2026-09-09): You've hit your session limit · resets 1pm (America/Los_Angeles) — https://github.com/anthropics/claude-code/issues/93148 (verified 2026-09-17) 4. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 5. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 6. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 7. Claude Code docs: Manage costs effectively (/usage, shared limits) — https://code.claude.com/docs/en/costs (verified 2026-09-18) ### 「You've hit your session limit · resets {time}」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-you-have-hit-your-session-limit-resets Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your session limit · resets {time} Variants: You've hit your session limit · resets 3:45pm | You've hit your session limit · resets 1pm (America/Los_Angeles) 「You've hit your session limit · resets {time}」是 Claude Code 在会话额度用完时的提示。Anthropic 说明会话额度每 5 小时刷新，文案里的时间是你自己账号的时钟，/usage 能看到。v2.1.234 起可原地等待并自动继续。它不是本站追踪的全员重置；板面行显示最近一次官方重置。 「You've hit your session limit · resets {time}」是什么意思？ 订阅计划自带滚动刷新的用量额度。会话部分用完时，Claude Code 显示这句并拦截后续请求，直到文案里的重置时间。官方 Error reference 把它和另外三句并列：周限额、Opus 限额、Sonnet 限额。 会话与周限额跨所有模型共享，用 /model 换模型无法恢复。文档还提醒：用量同时计入会话与周两个额度，一次大规模并行（workflow fanout）可能在会话窗口刷新前就把周额度打空。 v2.1.234 起，用 claude.ai 订阅登录的交互式会话可以原地等待：底部显示「Usage limit reached · continuing automatically at 3:45pm · esc to cancel」，刷新后任务从中断处继续。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage 看到它该怎么办 1. 先读文案里的时间: 那是你自己的刷新时刻，不是全站事件。等到那个时间，或者把会话开着让 Claude Code 替你等。 2. 打开自动继续: CLI 在 /config 里打开「Continue automatically at usage limit」；Desktop 的 Code 标签里，会话限额卡片有「Auto-continue when limits reset」勾选框。两者是独立设置。 3. 运行 /usage: 它显示各窗口的用量与刷新时间，能分清撞的是会话限额还是周限额。 4. 等不及就买额外用量: Pro / Max 运行 /usage-credits 开通 usage credits；Team / Enterprise 用它向管理员申请。开通后重置时间不变。 5. 升级提高基础额度: 各计划额度见 claude.com/pricing；Max 5x 与 Max 20x 分别是 Pro 每会话额度的 5 倍与 20 倍。 什么时候重置？ 你自己的窗口：Anthropic 写明会话额度每 5 小时刷新，具体时刻就是文案与 /usage 里的那个时间。usage credits 不改变它：「Usage credits don't affect this reset timing.」可以用下方链接的窗口倒计时把这个时间记下来。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次；订阅提醒可以在公告出现时收到通知。 FAQ: Q: 换模型能解决「You've hit your session limit」吗？ A: 不能。Error reference 写明会话与周限额跨所有模型共享，换模型不会恢复。只有按模型家族计的两句（「You've hit your Opus limit」「You've hit your Sonnet limit」）可以用 /model 切到家族外的模型绕开。 Q: 刷新后 Claude Code 会自己继续吗？ A: 会，v2.1.234 起，条件是用 claude.ai 订阅登录的交互式会话。等待时底部显示「Usage limit reached · continuing automatically at 3:45pm · esc to cancel」，在空提示符按 Esc 可取消；也可以在 /config 里关掉「Continue automatically at usage limit」。 Q: 会话限额和周限额是一回事吗？ A: 不是。会话限额每 5 小时刷新；周限额在分配给你账号的固定时刻每周刷新。用量同时计入两者，所以一次大并发可能在会话窗口刷新前先把周额度用光；/usage 两条都显示。 Q: 这和板上的官方 Claude 重置是一回事吗？ A: 不是。这句说的是你自己的 5 小时窗口。板面追踪的是 Anthropic 公开宣布的全员重置，每条链接原帖。就算上方板面行显示有重置记录，那也是与你文案里的时间无关的另一件事。 Sources: 1. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 2. Claude Code docs: Interactive mode — Wait for a usage limit to reset — https://code.claude.com/docs/en/interactive-mode#wait-for-a-usage-limit-to-reset (verified 2026-09-18) 3. anthropics/claude-code issue #93148 (2026-09-09): You've hit your session limit · resets 1pm (America/Los_Angeles) — https://github.com/anthropics/claude-code/issues/93148 (verified 2026-09-17) 4. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 5. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 6. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 7. Claude Code docs: Manage costs effectively (/usage, shared limits) — https://code.claude.com/docs/en/costs (verified 2026-09-18) ### "You've hit your weekly limit · resets {day} {time}" — Claude Code URL: https://didreset.com/errors/claude-code-you-have-hit-your-weekly-limit-resets Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your weekly limit · resets {day} {time} Variants: You've hit your weekly limit · resets Mon 12:00am | You've hit your weekly limit · resets Sep 18 at 3am (America/Los_Angeles) | You've hit your weekly limit · resets 4am (Europe/Madrid) | You've hit your weekly limit · resets 5pm (America/Detroit) "You've hit your weekly limit · resets {day} {time}" is Claude Code's message when your weekly allowance is used up. Anthropic says the weekly limit resets at a fixed time each week that is assigned to your account, and the message names that time; /usage shows it too. Only waiting, usage credits or an upgrade get you going earlier. It is not the account-wide reset this board tracks; the row below shows the last one on record. What does "You've hit your weekly limit · resets {day} {time}" mean? Pro and Max plans have a weekly usage limit on top of the five-hour session limit. It applies across all models, and Anthropic says it resets at a fixed time each week that is assigned to your account; your reset day and time stay the same regardless of when you start using Claude. The Error reference shows the message with an example of "resets Mon 12:00am". Real reports carry a date and a time zone, such as "resets Sep 18 at 3am (America/Los_Angeles)" or "resets 4am (Europe/Madrid)"; three issues with this exact title were opened between 2026-09-10 and 2026-09-16. The docs warn that usage counts against the session and weekly allowances at the same time, so a single burst such as a large workflow fanout can exhaust the weekly allowance before the session window resets. Weekly limits were announced on 2025-07-28 for late August 2025, estimated to affect less than 5% of subscribers. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai What to do when you see it 1. Run /usage: It shows the weekly bar and the exact reset time, so you can confirm it is the weekly window rather than the five-hour one. 2. Check Settings > Usage on claude.ai: The same page shows when the weekly limit resets for Opus only and for all other models. 3. Buy usage credits if you cannot wait: On Pro and Max, /usage-credits turns on usage credits; on Team and Enterprise it requests them from your admin. Credits do not move the reset time. 4. Upgrade for a bigger allowance: Max 5x and Max 20x include five and twenty times the Pro per-session allowance; the Error reference points to claude.com/pricing for higher base limits. 5. Do not expect auto-continue on Desktop: The Desktop app's session-limit card offers an "Auto-continue when limits reset" checkbox; the weekly-limit card does not. Plan the wait instead. When does it reset? Your own window: the weekly limit resets at a fixed time each week that is assigned to your account, which is the time in the message. Usage credits do not change it. Put that time on a countdown with the window tracker linked below; the five-hour session window keeps resetting on its own clock in the meantime, but it cannot lend you weekly capacity. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: Why did my reset time jump from tonight to next week? A: Because you crossed from the five-hour session limit into the weekly limit. The session window resets every five hours, but the weekly window resets only at the fixed weekly time assigned to your account. /usage shows both bars, so you can see which one is at 100%. Q: Does the weekly limit reset at the same time every week? A: Yes. Anthropic's help center says the weekly limit "resets at a fixed time each week that is assigned to your account", and that your reset day and time stay the same regardless of when you start using Claude or when your subscription begins. Q: Can Claude Code wait and continue automatically after a weekly reset? A: In the CLI, the automatic wait added in v2.1.234 applies to usage limits in an interactive subscription session. In the Desktop app the docs are explicit that the weekly-limit card has no "Auto-continue when limits reset" checkbox, only the session-limit card does. Q: How much is the weekly limit? A: Not published. Anthropic states that Max 5x and Max 20x include five and twenty times the Pro per-session allowance and that weekly limits were expected to affect less than 5% of subscribers when announced on 2025-07-28, but it does not publish a token or message count. This site does not estimate one. Sources: 1. anthropics/claude-code issue #94867 (2026-09-16): You've hit your weekly limit · resets Sep 18 at 3am (America/Los_Angeles) — https://github.com/anthropics/claude-code/issues/94867 (verified 2026-09-17) 2. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 3. anthropics/claude-code issue #93340 (2026-09-10): You've hit your weekly limit · resets 4am (Europe/Madrid) — https://github.com/anthropics/claude-code/issues/93340 (verified 2026-09-17) 4. anthropics/claude-code issue #94390 (2026-09-14): You've hit your weekly limit · resets 5pm (America/Detroit) — https://github.com/anthropics/claude-code/issues/94390 (verified 2026-09-17) 5. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 6. Claude Help Center: What is the Pro plan? — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 7. Anthropic on X, 2025-07-28: weekly rate limits for Pro and Max — https://x.com/AnthropicAI/status/1949898502688903593 (verified 2026-09-17) 8. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 9. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) ### 「You've hit your weekly limit · resets {day} {time}」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-you-have-hit-your-weekly-limit-resets Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your weekly limit · resets {day} {time} Variants: You've hit your weekly limit · resets Mon 12:00am | You've hit your weekly limit · resets Sep 18 at 3am (America/Los_Angeles) | You've hit your weekly limit · resets 4am (Europe/Madrid) | You've hit your weekly limit · resets 5pm (America/Detroit) 「You've hit your weekly limit · resets {day} {time}」是 Claude Code 在周额度用完时的提示。Anthropic 说明周限额每周在分配给你账号的固定时刻刷新，文案里写的就是那个时间，/usage 也能看到。只有等待、usage credits 或升级能提前继续。它不是本站追踪的全员重置。 「You've hit your weekly limit · resets {day} {time}」是什么意思？ Pro 与 Max 计划在 5 小时会话限额之上还有周限额。它跨所有模型计算，Anthropic 写明它在分配给你账号的固定时刻每周刷新；无论你什么时候开始用 Claude，刷新的日子和时刻都不变。 Error reference 的示例是「resets Mon 12:00am」。真实报告带日期与时区，例如「resets Sep 18 at 3am (America/Los_Angeles)」「resets 4am (Europe/Madrid)」；2026-09-10 到 2026-09-16 一周内官方仓库就有三条以此为题的 issue。 文档提醒用量同时计入会话与周两个额度，一次大规模并行（workflow fanout）可能在会话窗口刷新前就把周额度打空。周限额于 2025-07-28 宣布、2025 年 8 月下旬上线，官方估计影响不到 5% 的订阅者。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage 看到它该怎么办 1. 运行 /usage: 它显示周进度条和准确的刷新时间，可以确认撞的是周窗口而不是 5 小时窗口。 2. 看 claude.ai 的 Settings > Usage: 同一页显示周限额分别对 Opus 与其他模型何时刷新。 3. 等不及就开通 usage credits: Pro / Max 运行 /usage-credits 开通；Team / Enterprise 用它向管理员申请。开通后刷新时间不变。 4. 升级扩大额度: Max 5x 与 Max 20x 分别是 Pro 每会话额度的 5 倍与 20 倍；Error reference 指向 claude.com/pricing 查看更高的基础额度。 5. 别指望 Desktop 自动继续: Desktop 只有会话限额卡片有「Auto-continue when limits reset」勾选框，周限额卡片没有。只能安排好等待。 什么时候重置？ 你自己的窗口：周限额在分配给你账号的固定时刻每周刷新，就是文案里的时间，usage credits 不会改变它。可以用下方链接的窗口倒计时记下这个时间；这期间 5 小时会话窗口仍按自己的时钟刷新，但它借不出周额度。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: 为什么刷新时间从今晚跳到了下周？ A: 因为你从 5 小时会话限额撞进了周限额。会话窗口每 5 小时刷新，周窗口只在分配给你账号的固定周时刻刷新。/usage 两条进度条都显示，能看出哪一条到了 100%。 Q: 周限额每周都在同一时刻刷新吗？ A: 是。Anthropic 帮助中心写明周限额「resets at a fixed time each week that is assigned to your account」，而且无论你何时开始使用 Claude、订阅何时开始，刷新的日子和时刻都不变。 Q: 周限额刷新后 Claude Code 会自动继续吗？ A: CLI 里 v2.1.234 加入的自动等待适用于订阅登录的交互式会话遇到用量限额的情况。Desktop 则文档写得很明确：周限额卡片没有「Auto-continue when limits reset」勾选框，只有会话限额卡片有。 Q: 周限额到底是多少？ A: 未公布。Anthropic 只说明 Max 5x 与 Max 20x 是 Pro 每会话额度的 5 倍与 20 倍，以及 2025-07-28 宣布时预计周限额影响不到 5% 的订阅者，没有公布 token 或消息数。本站不估算。 Sources: 1. anthropics/claude-code issue #94867 (2026-09-16): You've hit your weekly limit · resets Sep 18 at 3am (America/Los_Angeles) — https://github.com/anthropics/claude-code/issues/94867 (verified 2026-09-17) 2. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 3. anthropics/claude-code issue #93340 (2026-09-10): You've hit your weekly limit · resets 4am (Europe/Madrid) — https://github.com/anthropics/claude-code/issues/93340 (verified 2026-09-17) 4. anthropics/claude-code issue #94390 (2026-09-14): You've hit your weekly limit · resets 5pm (America/Detroit) — https://github.com/anthropics/claude-code/issues/94390 (verified 2026-09-17) 5. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 6. Claude Help Center: What is the Pro plan? — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 7. Anthropic on X, 2025-07-28: weekly rate limits for Pro and Max — https://x.com/AnthropicAI/status/1949898502688903593 (verified 2026-09-17) 8. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 9. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) ### "You've hit your Opus limit · resets {time}" — Claude Code URL: https://didreset.com/errors/claude-code-you-have-hit-your-opus-limit-resets Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your Opus limit · resets {time} Variants: You've hit your Opus limit · resets 3:45pm | You've hit your Sonnet limit · resets 3:45pm | Claude Opus 4 limit reached, now using Sonnet 4 "You've hit your Opus limit · resets {time}" is Claude Code's message when the allowance for one model family is used up. The official Error reference says the Opus and Sonnet limits each apply only to requests to that family, so switching to a model outside the family with /model keeps you working; /usage shows when the limit resets. It is not the account-wide reset this board tracks; the row below shows the last one on record. What does "You've hit your Opus limit · resets {time}" mean? Besides the session and weekly limits that are shared across all models, Claude Code has limits per model family. When the Opus one is full you see "You've hit your Opus limit · resets {time}"; the Sonnet version reads "You've hit your Sonnet limit · resets 3:45pm" in the docs' example. The Error reference states that these limits each apply only to requests to that model family, so switching to a model outside the family with /model keeps you working. Each model has its own prompt cache, so the first request after the switch re-reads the whole conversation with no cache hits. This replaced an older behavior. In July 2025 Claude Code fell back silently with "Claude Opus 4 limit reached, now using Sonnet 4"; issue #3434 collected complaints from Max 20x users about that switch. claude.ai's Settings > Usage page still shows a separate weekly bar for Opus only. Which window is full? - Window: Opus limit (model family) · Resets: At the time in the message; applies only to requests to Opus models · Where to check: /usage in Claude Code; Settings > Usage on claude.ai shows the Opus-only weekly bar - Window: Sonnet limit (model family) · Resets: At the time in the message; applies only to requests to Sonnet models · Where to check: /usage in Claude Code - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai What to do when you see it 1. Switch families with /model: Pick a model outside the family named in the message. The docs say this keeps you working; the session and weekly limits are unaffected by the switch. 2. Expect one slow request: Each model has its own prompt cache, so the next request re-reads the whole conversation with no cache hits. That first turn draws more usage than usual. 3. Run /usage: It shows the Opus, session and weekly bars and when each resets, so you can decide whether to wait or keep going on another family. 4. Or wait for the time in the message: Only the Opus limit is blocked; you can keep the session open and switch back after the reset. When does it reset? Your own window: the message names the reset time for the Opus family, and /usage shows the same clock. claude.ai's Settings > Usage page lets you check when your plan's weekly limit resets for Opus only and for all other models. Put the time on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: Is the Opus limit separate from the session limit? A: Yes. The Error reference says the session and weekly limits are shared across all models, while the Opus and Sonnet limits each apply only to requests to that model family. That is why /model to another family works for the Opus message but not for the session or weekly one. Q: Do I lose my conversation when I switch to Sonnet? A: No. The conversation stays; what you lose is the prompt cache. Each model has its own cache, so the next request re-reads the whole conversation with no cache hits, which costs more usage for that one turn. Q: What did "Claude Opus 4 limit reached, now using Sonnet 4" mean? A: It was the July 2025 behavior: Claude Code switched to Sonnet on its own when the Opus allowance ran out. Issue #3434 tracked complaints from Max 20x users. Current versions stop with "You've hit your Opus limit" and let you choose the model with /model. Q: Where do I see the Opus limit on claude.ai? A: Anthropic's usage-limit best-practices article says Settings > Usage shows when your plan's weekly limit resets for Opus only and for all other models. In Claude Code, /usage shows the same information. Sources: 1. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 2. anthropics/claude-code issue #3434 (2025-07-13): Claude silently falls back to Sonnet after Opus cap (body quotes "Claude Opus 4 limit reached, now using Sonnet 4") — https://github.com/anthropics/claude-code/issues/3434 (verified 2026-09-17) 3. Claude Code docs: Manage costs effectively (/usage, shared limits) — https://code.claude.com/docs/en/costs (verified 2026-09-18) 4. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 5. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) ### 「You've hit your Opus limit · resets {time}」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-you-have-hit-your-opus-limit-resets Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your Opus limit · resets {time} Variants: You've hit your Opus limit · resets 3:45pm | You've hit your Sonnet limit · resets 3:45pm | Claude Opus 4 limit reached, now using Sonnet 4 「You've hit your Opus limit · resets {time}」是 Claude Code 在某一模型家族额度用完时的提示。Error reference 写明 Opus 与 Sonnet 限额只约束对该家族的请求，用 /model 切到家族外的模型即可继续；/usage 显示何时刷新。它不是本站追踪的全员重置。 「You've hit your Opus limit · resets {time}」是什么意思？ 除了跨所有模型共享的会话与周限额，Claude Code 还有按模型家族计的限额。Opus 那条满了显示「You've hit your Opus limit · resets {time}」；Sonnet 版本在文档示例里是「You've hit your Sonnet limit · resets 3:45pm」。 Error reference 写明这类限额只约束对该家族的请求，用 /model 切到家族外的模型可以继续工作。每个模型有自己的 prompt cache，切换后的第一次请求会重读整个对话、没有缓存命中。 这取代了旧行为。2025 年 7 月 Claude Code 会静默降级并提示「Claude Opus 4 limit reached, now using Sonnet 4」，issue #3434 汇集了 Max 20x 用户对这种切换的投诉。claude.ai 的 Settings > Usage 至今仍单独显示一条 Opus 专用的周进度条。 满的是哪个窗口？ - 窗口: Opus 限额（模型家族） · 何时刷新: 文案里的时间；只约束对 Opus 模型的请求 · 在哪里看: Claude Code 里 /usage；claude.ai 的 Settings > Usage 有 Opus 专用周进度条 - 窗口: Sonnet 限额（模型家族） · 何时刷新: 文案里的时间；只约束对 Sonnet 模型的请求 · 在哪里看: Claude Code 里 /usage - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage 看到它该怎么办 1. 用 /model 换一个家族: 选文案里那个家族之外的模型。文档说这样可以继续工作；会话与周限额不受切换影响。 2. 接受一次慢请求: 每个模型有自己的 prompt cache，切换后的下一次请求会重读整个对话、没有缓存命中，那一轮用量会比平时多。 3. 运行 /usage: 它显示 Opus、会话、周三条进度条和各自的刷新时间，方便决定是等还是换家族继续。 4. 或者等文案里的时间: 被拦的只有 Opus 限额；可以把会话开着，刷新后再切回来。 什么时候重置？ 你自己的窗口：文案里写的是 Opus 家族的刷新时间，/usage 显示同一个时钟。claude.ai 的 Settings > Usage 可以分别查看周限额对 Opus 与其他模型何时刷新。可以用下方链接的窗口倒计时记下这个时间。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: Opus 限额和会话限额是分开的吗？ A: 是。Error reference 写明会话与周限额跨所有模型共享，而 Opus 与 Sonnet 限额各自只约束对该家族的请求。所以 /model 换家族对 Opus 这句有效，对会话或周限额那句无效。 Q: 切到 Sonnet 会丢掉对话吗？ A: 不会。对话还在，丢的是 prompt cache。每个模型有自己的缓存，切换后的下一次请求会重读整个对话、没有缓存命中，那一轮会多花一些用量。 Q: 「Claude Opus 4 limit reached, now using Sonnet 4」是什么意思？ A: 那是 2025 年 7 月的行为：Opus 额度用完后 Claude Code 自己切到 Sonnet。issue #3434 记录了 Max 20x 用户的投诉。现在的版本会停在「You've hit your Opus limit」，由你用 /model 决定换哪个模型。 Q: 在 claude.ai 上哪里能看 Opus 限额？ A: Anthropic 的额度最佳实践文章写明，Settings > Usage 会分别显示周限额对 Opus 与其他模型何时刷新。在 Claude Code 里 /usage 显示同样的信息。 Sources: 1. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 2. anthropics/claude-code issue #3434 (2025-07-13): Claude silently falls back to Sonnet after Opus cap (body quotes "Claude Opus 4 limit reached, now using Sonnet 4") — https://github.com/anthropics/claude-code/issues/3434 (verified 2026-09-17) 3. Claude Code docs: Manage costs effectively (/usage, shared limits) — https://code.claude.com/docs/en/costs (verified 2026-09-18) 4. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 5. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) ### "Approaching weekly usage limit — Resets {Day}, {Mon D}, {H:MM PM}" — Claude Code URL: https://didreset.com/errors/claude-code-approaching-weekly-usage-limit-banner Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Approaching weekly usage limit — Resets {Day}, {Mon D}, {H:MM PM} Variants: Approaching weekly usage limit — Resets Sun, Sep 20, 2:00 PM | You've used 85% of your session limit · resets 3:45pm "Approaching weekly usage limit — Resets {Day}, {Mon D}, {H:MM PM}" is a warning banner above the Claude Code input, not a block: you can keep working. The Error reference describes the usual warning with a percentage, such as "You've used 85% of your session limit · resets 3:45pm". Issue #94694 reports the percentage-free banner at 54% of the weekly limit on CLI 2.1.271. The reset in the banner is your own weekly time; /usage shows the real number. What does "Approaching weekly usage limit — Resets {Day}, {Mon D}, {H:MM PM}" mean? Before a window runs out, Claude Code can warn you that you have used most of it. The Error reference gives the form "You've used 85% of your session limit · resets 3:45pm". The weekly banner above the prompt reads "Approaching weekly usage limit — Resets Sun, Sep 20, 2:00 PM" in issue #94694 and carries no percentage. Issue #94694 (2026-09-16, CLI 2.1.271 and Desktop 2.110.0, Max 20x) reports the banner appearing at 54% of the weekly limit. The reporter's explanation, that the server sends a threshold flag without a utilization value, is their reading of the client code, not an official statement; the issue is labeled as a duplicate. The date and time after "Resets" are your own weekly reset: Anthropic says the weekly limit resets at a fixed time each week that is assigned to your account. The banner is not the account-wide reset this board tracks. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai What to do when you see it 1. Keep working: The banner is a warning. Requests are blocked only when you see "You've hit your weekly limit"; until then nothing has changed. 2. Check the real percentage: Run /usage, or in the Desktop app click the usage ring next to the model picker. Issue #94694 saw the banner at 54%, so the banner alone does not tell you how close you are. 3. Watch it continuously if you prefer: The Error reference suggests adding the rate_limits fields to a custom status line to see remaining allowance without banners. 4. Plan around the reset time: The time after "Resets" is fixed for your account each week. If a large fanout is coming, remember usage counts against the session and weekly allowances at the same time. When does it reset? Your own window: the banner's "Resets" time is when your weekly allowance refreshes, at a fixed time each week that is assigned to your account. Settings > Usage on claude.ai shows the same reset separately for Opus and for all other models. Put it on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: Does the banner mean I am blocked? A: No. It is a warning that you are approaching the weekly limit. Claude Code blocks requests only when the window is actually used up, which shows as "You've hit your weekly limit · resets …". Until then you can keep sending prompts. Q: Why did it appear at 54%? A: Issue #94694 reports exactly that on CLI 2.1.271 and Desktop 2.110.0. The reporter attributes it to a threshold flag sent without a utilization value, but that is their reading of the client code, not an official explanation. Run /usage for the real number. Q: How is this different from "You've used 85% of your session limit"? A: That is the form the Error reference documents: a percentage plus the reset time for the session window. The weekly banner in #94694 has the same purpose for the weekly window but shows no percentage, only the reset date and time. Q: Can I hide the banner? A: The Error reference describes the warning and points to the rate_limits status-line fields and the Desktop usage ring for watching usage; it does not describe a setting to hide the banner. Issue #94694 is the place to follow that request. Sources: 1. anthropics/claude-code issue #94694 (2026-09-16): "Approaching weekly usage limit" banner fires at 54% weekly usage (CLI 2.1.271 / Desktop 2.110.0, Max 20x) — https://github.com/anthropics/claude-code/issues/94694 (verified 2026-09-17) 2. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 3. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 4. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) ### 「Approaching weekly usage limit — Resets {Day}, {Mon D}, {H:MM PM}」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-approaching-weekly-usage-limit-banner Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Approaching weekly usage limit — Resets {Day}, {Mon D}, {H:MM PM} Variants: Approaching weekly usage limit — Resets Sun, Sep 20, 2:00 PM | You've used 85% of your session limit · resets 3:45pm 「Approaching weekly usage limit — Resets …」是 Claude Code 输入框上方的预警横幅，不是拦截，可以继续工作。issue #94694 报告它在周限额 54% 时就出现；Error reference 记录的常规预警带百分比。Resets 是你自己的周刷新时刻，真实数字看 /usage。 「Approaching weekly usage limit — Resets {Day}, {Mon D}, {H:MM PM}」是什么意思？ 窗口快用完之前，Claude Code 会提醒你已经用掉大部分。Error reference 给的形式是「You've used 85% of your session limit · resets 3:45pm」。输入框上方的周横幅在 issue #94694 里写作「Approaching weekly usage limit — Resets Sun, Sep 20, 2:00 PM」，不带百分比。 issue #94694（2026-09-16，CLI 2.1.271 与 Desktop 2.110.0，Max 20x）报告横幅在周限额 54% 时出现。报告者的解释是服务端只发阈值标记、不带 utilization 数值，那是他对客户端代码的解读，不是官方说法；该 issue 被标为 duplicate。 「Resets」后面的日期和时间是你自己的周刷新时刻：Anthropic 写明周限额在分配给你账号的固定时刻每周刷新。横幅不是本站追踪的全员重置。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage 看到它该怎么办 1. 继续工作: 横幅只是预警。只有看到「You've hit your weekly limit」才是真的被拦；在那之前什么都没变。 2. 看真实百分比: 运行 /usage，或在 Desktop 点模型选择器旁边的用量环。issue #94694 在 54% 就看到了横幅，所以光看横幅判断不了离上限多近。 3. 想持续观察就用状态栏: Error reference 建议把 rate_limits 字段加进自定义 status line，不靠横幅也能看到剩余额度。 4. 按刷新时间安排工作: 「Resets」后面的时间每周对你的账号是固定的。如果要跑大规模并行，记得用量同时计入会话与周两个额度。 什么时候重置？ 你自己的窗口：横幅里的「Resets」时间就是你周额度刷新的时刻，每周在分配给你账号的固定时刻。claude.ai 的 Settings > Usage 会分别显示 Opus 与其他模型的周刷新时间。可以用下方链接的窗口倒计时记下它。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: 出现横幅代表被拦了吗？ A: 不是。它只是提醒你接近周限额。Claude Code 只在窗口真正用完时拦截请求，那时显示的是「You've hit your weekly limit · resets …」。在那之前可以继续发提示。 Q: 为什么 54% 就出现了？ A: issue #94694 在 CLI 2.1.271 与 Desktop 2.110.0 上报告的正是这个现象。报告者归因于服务端发了不带 utilization 数值的阈值标记，但那是他对客户端代码的解读，不是官方解释。真实数字运行 /usage 看。 Q: 它和「You've used 85% of your session limit」有什么区别？ A: 后者是 Error reference 记录的形式：百分比加会话窗口的刷新时间。#94694 里的周横幅作用相同、针对周窗口，但不显示百分比，只有刷新的日期与时间。 Q: 能把横幅关掉吗？ A: Error reference 描述了这个预警，并指出可以用 status line 的 rate_limits 字段或 Desktop 的用量环观察用量；它没有描述隐藏横幅的设置。这个诉求可以在 issue #94694 里跟进。 Sources: 1. anthropics/claude-code issue #94694 (2026-09-16): "Approaching weekly usage limit" banner fires at 54% weekly usage (CLI 2.1.271 / Desktop 2.110.0, Max 20x) — https://github.com/anthropics/claude-code/issues/94694 (verified 2026-09-17) 2. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 3. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 4. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) ### "You've hit your limit · resets {time} ({Region/City})" — Claude Code URL: https://didreset.com/errors/claude-code-you-have-hit-your-limit-resets-timezone Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your limit · resets {time} ({Region/City}) Variants: You've hit your limit · resets Feb 20, 5pm (Africa/Libreville) | You've hit your limit · resets 2pm (UTC) | You've hit your limit · resets 2am (Europe/Berlin) "You've hit your limit · resets {time} ({Region/City})" is the Claude Code wording from early 2026 (v2.1.37 to v2.1.41 in the reports) that did not say whether the five-hour session or the weekly window was full. The time and IANA zone in the brackets are your own reset time; /usage tells you which window it is. Current versions split it into session, weekly, Opus and Sonnet messages. It is not the account-wide reset this board tracks. What does "You've hit your limit · resets {time} ({Region/City})" mean? In January and February 2026, Claude Code showed one generic limit message with a time and an IANA time zone in brackets, such as "resets 2pm (UTC)" or "resets 2am (Europe/Berlin)". Several issues with this exact title were opened in that period (#25607, #24428, #21111 among them). The message did not say which window was full. In #25607 the reset time jumped from "10pm today" to seven days later while /usage showed 12%, which is what running into the weekly window looks like. In #24428 the underlying error was an API 429 rate_limit_error: "This request would exceed your account's rate limit. Please try again later." The current Error reference no longer lists this form. It shows four messages instead: "You've hit your session limit", "You've hit your weekly limit", "You've hit your Opus limit" and "You've hit your Sonnet limit", so upgrading Claude Code tells you which window you hit. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai What to do when you see it 1. Run /usage: It shows the session and weekly bars and when each resets. If the reset in the message is days away, it is the weekly window; if it is within five hours, it is the session window. 2. Convert the time zone: The zone in brackets is an IANA name such as Europe/Berlin or UTC. If it is not your local zone, convert the time before deciding how long to wait. 3. Confirm which account is signed in: If claude.ai still answers for the same account, run /status to see the active credential and /login to switch to the subscription you expect, as the help center describes. 4. Update Claude Code: Current versions replace this line with a message that names the window (session, weekly, Opus or Sonnet) and, since v2.1.234, can wait and continue automatically. 5. Report it if the numbers disagree: If /usage shows plenty of room and the block persists, open an issue on the anthropics/claude-code repository with the message and the /usage output. When does it reset? Your own window: the time in brackets is when your window refreshes. The session limit resets every five hours; the weekly limit resets at a fixed time each week that is assigned to your account. Put the time on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: Is "You've hit your limit" the session limit or the weekly limit? A: The old wording did not say. Run /usage: it shows both bars and their reset times. A reset several days away means the weekly window; a reset within five hours means the session window. Current Claude Code versions name the window in the message itself. Q: Why did the reset time jump from tonight to next week? A: That is what crossing into the weekly window looks like. Issue #25607 reported the time moving from "10pm today" to seven days later while /usage showed 12%. The weekly limit resets at a fixed time each week that is assigned to your account. Q: What is the error underneath? A: In issue #24428 the API returned a 429 rate_limit_error with the text "This request would exceed your account's rate limit. Please try again later." The Claude Platform error reference lists 429 rate_limit_error as the code for hitting a rate limit. Q: Does the time zone in brackets change the reset? A: No, it only tells you how to read the time. Claude Code prints the IANA zone it used, for example (UTC) or (Africa/Libreville). Convert to your local zone; the underlying reset moment is the same. Sources: 1. anthropics/claude-code issue #25607 (2026-02-13): You've hit your limit · resets Feb 20, 5pm (Africa/Libreville) — https://github.com/anthropics/claude-code/issues/25607 (verified 2026-09-17) 2. anthropics/claude-code issue #24428 (2026-02-09): You've hit your limit · resets 2pm (UTC) — https://github.com/anthropics/claude-code/issues/24428 (verified 2026-09-17) 3. anthropics/claude-code issue #21111 (2026-01-26): You've hit your limit · resets 2am (Europe/Berlin) — https://github.com/anthropics/claude-code/issues/21111 (verified 2026-09-17) 4. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 5. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) 6. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 7. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 8. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) ### 「You've hit your limit · resets {time} ({Region/City})」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-you-have-hit-your-limit-resets-timezone Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your limit · resets {time} ({Region/City}) Variants: You've hit your limit · resets Feb 20, 5pm (Africa/Libreville) | You've hit your limit · resets 2pm (UTC) | You've hit your limit · resets 2am (Europe/Berlin) 「You've hit your limit · resets {time} ({Region/City})」是 Claude Code 2026 年初（v2.1.37 到 v2.1.41）的文案，不说明满的是会话窗口还是周窗口。括号里的时间与时区是你自己的刷新时刻，/usage 能看出是哪个窗口。现在的版本已拆成四句写明窗口。 「You've hit your limit · resets {time} ({Region/City})」是什么意思？ 2026 年 1 月到 2 月，Claude Code 只显示一句通用的限额提示，带时间和括号里的 IANA 时区，例如「resets 2pm (UTC)」「resets 2am (Europe/Berlin)」。那段时间官方仓库出现多条以此为题的 issue（#25607、#24428、#21111 等）。 这句不说明满的是哪个窗口。#25607 里刷新时间从「10pm today」跳到 7 天后、/usage 显示 12%，这正是撞进周窗口的样子。#24428 底层是 API 的 429 rate_limit_error：「This request would exceed your account's rate limit. Please try again later.」 现在的 Error reference 已不再列这种写法，而是四句：「You've hit your session limit」「You've hit your weekly limit」「You've hit your Opus limit」「You've hit your Sonnet limit」。升级 Claude Code 就能直接看到撞的是哪个窗口。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage 看到它该怎么办 1. 运行 /usage: 它显示会话与周两条进度条和各自的刷新时间。文案里的刷新在几天之后就是周窗口；在 5 小时之内就是会话窗口。 2. 换算时区: 括号里是 IANA 时区名，例如 Europe/Berlin 或 UTC。如果不是你所在的时区，先换算再决定等多久。 3. 确认登录的是哪个账号: 如果同一账号在 claude.ai 上还能用，运行 /status 看当前生效的凭证，按帮助中心的说明用 /login 切换到你想用的订阅。 4. 更新 Claude Code: 现在的版本会把这句换成写明窗口（会话、周、Opus 或 Sonnet）的提示，并且从 v2.1.234 起可以原地等待、自动继续。 5. 数字对不上就报 issue: 如果 /usage 显示还有很多余量却持续被拦，带上报错原文和 /usage 输出到 anthropics/claude-code 仓库提 issue。 什么时候重置？ 你自己的窗口：括号里的时间就是你窗口刷新的时刻。会话限额每 5 小时刷新；周限额在分配给你账号的固定时刻每周刷新。可以用下方链接的窗口倒计时记下这个时间。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: 「You've hit your limit」是会话限额还是周限额？ A: 旧文案不说明。运行 /usage：两条进度条和刷新时间都在。刷新在几天后的是周窗口，在 5 小时内的是会话窗口。现在的 Claude Code 版本会在文案里直接写明窗口。 Q: 为什么刷新时间从今晚跳到了下周？ A: 那就是撞进周窗口的样子。issue #25607 报告时间从「10pm today」变成 7 天后，而 /usage 显示 12%。周限额在分配给你账号的固定时刻每周刷新。 Q: 底层是什么错误？ A: issue #24428 里 API 返回 429 rate_limit_error，文本是「This request would exceed your account's rate limit. Please try again later.」。Claude Platform 的错误码文档把 429 rate_limit_error 列为触发速率限制的代码。 Q: 括号里的时区会改变刷新时间吗？ A: 不会，它只告诉你怎么读那个时间。Claude Code 打印的是它使用的 IANA 时区，例如 (UTC) 或 (Africa/Libreville)。换算成你的本地时区即可，底层的刷新时刻是同一个。 Sources: 1. anthropics/claude-code issue #25607 (2026-02-13): You've hit your limit · resets Feb 20, 5pm (Africa/Libreville) — https://github.com/anthropics/claude-code/issues/25607 (verified 2026-09-17) 2. anthropics/claude-code issue #24428 (2026-02-09): You've hit your limit · resets 2pm (UTC) — https://github.com/anthropics/claude-code/issues/24428 (verified 2026-09-17) 3. anthropics/claude-code issue #21111 (2026-01-26): You've hit your limit · resets 2am (Europe/Berlin) — https://github.com/anthropics/claude-code/issues/21111 (verified 2026-09-17) 4. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 5. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) 6. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 7. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 8. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) ### "Claude usage limit reached. Your limit will reset at {time} ({Region/City})." — Claude Code URL: https://didreset.com/errors/claude-usage-limit-reached-your-limit-will-reset-at Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Claude usage limit reached. Your limit will reset at {time} ({Region/City}). Variants: Claude usage limit reached. Your limit will reset at 3pm (America/Santiago). | Claude usage limit reached. Your limit will reset at 2pm (America/New_York) | Claude usage limit reached. Your limit will reset at 5pm (Europe/Warsaw). "Claude usage limit reached. Your limit will reset at {time} ({Region/City})." is the 2025 Claude Code wording (v1.0.x in the reports) for a used-up usage window. The time and the IANA zone in brackets are your own reset time; Anthropic says the session limit resets every five hours and the weekly limit at a fixed weekly time. Current versions name the window instead and, since v2.1.234, can wait automatically. It is not the account-wide reset this board tracks. What does "Claude usage limit reached. Your limit will reset at {time} ({Region/City})." mean? Through the second half of 2025, Claude Code reported a full usage window with this sentence, for example "Claude usage limit reached. Your limit will reset at 3pm (America/Santiago)." It did not say whether the five-hour session or the weekly window was full. Two issues show its rough edges. In #9236 (v1.0.39) the reset time was 23 hours away instead of the expected five-hour cycle. In #5977 Claude Code did not continue after the reset time passed; the automatic wait that fixes this arrived in v2.1.234. The current Error reference lists four messages instead: "You've hit your session limit", "You've hit your weekly limit", "You've hit your Opus limit" and "You've hit your Sonnet limit". If you still see the 2025 sentence, you are on an old build. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai What to do when you see it 1. Read the time and zone: The zone in brackets is an IANA name such as America/Santiago. Convert it to your local time before deciding how long to wait. 2. Run /usage: It shows the session and weekly bars and when each resets. A reset within five hours is the session window; days away is the weekly window. 3. Update Claude Code: Current versions name the window in the message, and since v2.1.234 an interactive subscription session can wait and continue automatically, showing "Usage limit reached · continuing automatically at 3:45pm · esc to cancel". 4. Wait, then send a new prompt: On the 2025 builds nothing resumes on its own after the reset (issue #5977); send the next prompt yourself once the time has passed. When does it reset? Your own window: the time in the message is when your window refreshes. Anthropic states the session limit resets every five hours and the weekly limit at a fixed time each week assigned to your account; issue #9236 shows the displayed time can be later than the five-hour cycle. Put the time on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: Why is the reset time more than five hours away? A: Either you hit the weekly window, which resets at a fixed weekly time assigned to your account, or the display was off: issue #9236 recorded a 23-hour gap on v1.0.39. Run /usage to see both bars and their reset times. Q: Will Claude Code continue on its own after the reset? A: Not on the builds that print this sentence; issue #5977 is exactly that complaint. Since v2.1.234, an interactive session signed in with a claude.ai subscription can wait in place and continue automatically, and the CLI setting "Continue automatically at usage limit" in /config controls it. Q: Is this message still current? A: No. The Error reference now lists "You've hit your session limit", "You've hit your weekly limit", "You've hit your Opus limit" and "You've hit your Sonnet limit". The older sentence appears in 2025 reports such as #9236 and #5977. Q: How do I read the time zone in brackets? A: It is the IANA zone Claude Code used, such as (America/New_York) or (Europe/Warsaw). The moment of the reset is the same everywhere; only the printed clock time depends on the zone, so convert it to yours. Sources: 1. anthropics/claude-code issue #9236 (2025-10-09): [BUG] Claude usage limit reached. Your limit will reset at 3pm (body: America/Santiago, v1.0.39) — https://github.com/anthropics/claude-code/issues/9236 (verified 2026-09-17) 2. anthropics/claude-code issue #5977 (2025-08-17): [BUG] Claude does not continue after "Claude usage limit reached. Your limit will reset at 2pm (America/New_York)" — https://github.com/anthropics/claude-code/issues/5977 (verified 2026-09-17) 3. anthropics/claude-code issue #6119 (2025-08-19): Claude AI usage limit reached|1755615600 (body shows the Europe/Warsaw form) — https://github.com/anthropics/claude-code/issues/6119 (verified 2026-09-17) 4. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 5. Claude Code docs: Interactive mode — Wait for a usage limit to reset — https://code.claude.com/docs/en/interactive-mode#wait-for-a-usage-limit-to-reset (verified 2026-09-18) 6. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 7. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) ### 「Claude usage limit reached. Your limit will reset at {time} ({Region/City}).」 — Claude Code URL: https://didreset.com/zh/errors/claude-usage-limit-reached-your-limit-will-reset-at Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Claude usage limit reached. Your limit will reset at {time} ({Region/City}). Variants: Claude usage limit reached. Your limit will reset at 3pm (America/Santiago). | Claude usage limit reached. Your limit will reset at 2pm (America/New_York) | Claude usage limit reached. Your limit will reset at 5pm (Europe/Warsaw). 「Claude usage limit reached. Your limit will reset at {time} …」是 Claude Code 2025 年（v1.0.x）的窗口用尽文案。括号里的时间与时区是你自己的刷新时刻；Anthropic 说明会话限额每 5 小时刷新、周限额在固定周时刻刷新。现在的版本已写明窗口并可自动等待。 「Claude usage limit reached. Your limit will reset at {time} ({Region/City}).」是什么意思？ 2025 年下半年，Claude Code 用这句报告用量窗口用尽，例如「Claude usage limit reached. Your limit will reset at 3pm (America/Santiago).」。它不说明满的是 5 小时会话窗口还是周窗口。 两条 issue 暴露了它的毛病：#9236（v1.0.39）里刷新时间距当时 23 小时，而不是预期的 5 小时周期；#5977 里到了刷新时间 Claude Code 也不会自己继续，这个问题要到 v2.1.234 的自动等待才解决。 现在的 Error reference 列的是四句：「You've hit your session limit」「You've hit your weekly limit」「You've hit your Opus limit」「You've hit your Sonnet limit」。如果你还看到 2025 年这句，说明用的是旧版本。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage 看到它该怎么办 1. 读时间和时区: 括号里是 IANA 时区名，例如 America/Santiago。先换算成本地时间再决定等多久。 2. 运行 /usage: 它显示会话与周两条进度条和各自的刷新时间。5 小时内刷新的是会话窗口，几天后刷新的是周窗口。 3. 更新 Claude Code: 现在的版本会在文案里写明窗口，并且从 v2.1.234 起订阅登录的交互式会话可以原地等待、自动继续，显示「Usage limit reached · continuing automatically at 3:45pm · esc to cancel」。 4. 等到时间后自己发下一条: 打印这句的 2025 年版本在刷新后不会自动恢复（issue #5977）；时间过了要自己发下一条提示。 什么时候重置？ 你自己的窗口：文案里的时间就是你窗口刷新的时刻。Anthropic 写明会话限额每 5 小时刷新、周限额在分配给你账号的固定时刻每周刷新；issue #9236 说明显示的时间可能比 5 小时周期晚。可以用下方链接的窗口倒计时记下这个时间。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: 为什么刷新时间在 5 小时之后？ A: 要么你撞的是周窗口，它在分配给你账号的固定周时刻刷新；要么显示有偏差：issue #9236 在 v1.0.39 上记录了 23 小时的间隔。运行 /usage 看两条进度条和各自的刷新时间。 Q: 刷新后 Claude Code 会自己继续吗？ A: 打印这句的版本不会，issue #5977 抱怨的正是这一点。v2.1.234 起，用 claude.ai 订阅登录的交互式会话可以原地等待并自动继续，由 /config 里的「Continue automatically at usage limit」控制。 Q: 这句文案还在用吗？ A: 不在了。Error reference 现在列的是「You've hit your session limit」「You've hit your weekly limit」「You've hit your Opus limit」「You've hit your Sonnet limit」。旧句子见于 2025 年的报告，例如 #9236 与 #5977。 Q: 括号里的时区怎么读？ A: 那是 Claude Code 使用的 IANA 时区，例如 (America/New_York) 或 (Europe/Warsaw)。刷新的时刻在哪里都一样，只是打印出来的钟点取决于时区，换算成你的时区即可。 Sources: 1. anthropics/claude-code issue #9236 (2025-10-09): [BUG] Claude usage limit reached. Your limit will reset at 3pm (body: America/Santiago, v1.0.39) — https://github.com/anthropics/claude-code/issues/9236 (verified 2026-09-17) 2. anthropics/claude-code issue #5977 (2025-08-17): [BUG] Claude does not continue after "Claude usage limit reached. Your limit will reset at 2pm (America/New_York)" — https://github.com/anthropics/claude-code/issues/5977 (verified 2026-09-17) 3. anthropics/claude-code issue #6119 (2025-08-19): Claude AI usage limit reached|1755615600 (body shows the Europe/Warsaw form) — https://github.com/anthropics/claude-code/issues/6119 (verified 2026-09-17) 4. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 5. Claude Code docs: Interactive mode — Wait for a usage limit to reset — https://code.claude.com/docs/en/interactive-mode#wait-for-a-usage-limit-to-reset (verified 2026-09-18) 6. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 7. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) ### "Claude AI usage limit reached|{unix_timestamp}" — Claude Code URL: https://didreset.com/errors/claude-ai-usage-limit-reached-unix-timestamp Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Claude AI usage limit reached|{unix_timestamp} Variants: Claude AI usage limit reached|1755615600 "Claude AI usage limit reached|{unix_timestamp}" is a raw form of the Claude Code usage-limit error seen on 2025 builds such as v1.0.84: the text before the bar is the message, and the number after it is your reset time in Unix seconds. In issue #6119 the value 1755615600 is 2025-08-19 15:00 UTC. Underneath is an API 429 rate_limit_error. Run /usage to see the same time; it is not the account-wide reset this board tracks. What does "Claude AI usage limit reached|{unix_timestamp}" mean? On some 2025 builds, Claude Code printed the usage-limit error exactly as it came from the API: a message, a vertical bar, and a Unix timestamp. Issue #6119 (v1.0.84) has "Claude AI usage limit reached|1755615600" as its title, and 1755615600 seconds since 1970 is 2025-08-19 15:00 UTC. The same issue shows the human-readable sentence in its body, "Claude usage limit reached. Your limit will reset at 5pm (Europe/Warsaw).", and the errors array carries an API 429 rate_limit_error: "This request would exceed your account's rate limit. Please try again later." The current Error reference lists formatted messages instead: "You've hit your session limit · resets 3:45pm" and its weekly, Opus and Sonnet siblings. The window behind them is the same: a five-hour session limit and a weekly limit that resets at a fixed time assigned to your account. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: /usage in Claude Code, or Settings > Usage on claude.ai What to do when you see it 1. Convert the number after the bar: It is seconds since 1970-01-01 UTC. Paste it into any epoch converter, or run date -r 1755615600 on macOS, to get the local reset time. 2. Run /usage: It shows the session and weekly bars with their reset times, so you can confirm which window the timestamp refers to. 3. Wait, then retry: Requests are blocked until the timestamp. On 2025 builds nothing resumes on its own; send the next prompt yourself. 4. Update Claude Code: Current versions print the reset time in words and, since v2.1.234, can wait and continue automatically in an interactive subscription session. 5. Report it if usage looks fine: If /usage shows room and the error keeps coming back, open an issue on anthropics/claude-code with the message and timestamp. When does it reset? Your own window: the timestamp is the exact reset moment for your window. Anthropic says the session limit resets every five hours and the weekly limit at a fixed time each week assigned to your account. Put the converted time on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: What does the number after the bar mean? A: It is a Unix timestamp: seconds since 1970-01-01 00:00 UTC. In issue #6119 the value 1755615600 converts to 2025-08-19 15:00 UTC, which is the moment that account's window refreshed. Convert it to your local time to know how long to wait. Q: What error is behind this message? A: An API 429 rate_limit_error. Issue #6119's errors array shows "This request would exceed your account's rate limit. Please try again later.", and the Claude Platform error reference lists 429 rate_limit_error as the code for hitting a rate limit. Q: Is this the five-hour window or the weekly window? A: The raw message does not say. Compare the timestamp with now: within five hours points to the session limit, days away points to the weekly limit. /usage shows both bars and their reset times directly. Q: Why do current versions not show this? A: The Error reference now documents formatted messages such as "You've hit your session limit · resets 3:45pm". The bar-and-timestamp form appears in 2025 reports on builds like v1.0.84. Sources: 1. anthropics/claude-code issue #6119 (2025-08-19): Claude AI usage limit reached|1755615600 (v1.0.84) — https://github.com/anthropics/claude-code/issues/6119 (verified 2026-09-17) 2. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) 3. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 4. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) ### 「Claude AI usage limit reached|{unix_timestamp}」 — Claude Code URL: https://didreset.com/zh/errors/claude-ai-usage-limit-reached-unix-timestamp Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Claude AI usage limit reached|{unix_timestamp} Variants: Claude AI usage limit reached|1755615600 「Claude AI usage limit reached|{unix_timestamp}」是 2025 年版本（如 v1.0.84）Claude Code 限额报错的原始形态：竖线后的数字是刷新时刻的 Unix 秒，issue #6119 的 1755615600 即 2025-08-19 15:00 UTC。底层是 429 错误。 「Claude AI usage limit reached|{unix_timestamp}」是什么意思？ 2025 年的部分版本会把限额错误按 API 返回的原样打印：文案、竖线、Unix 时间戳。issue #6119（v1.0.84）的标题就是「Claude AI usage limit reached|1755615600」，1755615600 秒即 2025-08-19 15:00 UTC。 同一条 issue 的正文里有可读的句子「Claude usage limit reached. Your limit will reset at 5pm (Europe/Warsaw).」，errors 数组里是 API 的 429 rate_limit_error：「This request would exceed your account's rate limit. Please try again later.」 现在的 Error reference 列的是格式化后的文案：「You've hit your session limit · resets 3:45pm」及其周、Opus、Sonnet 版本。背后的窗口没变：5 小时会话限额，加上在分配给账号的固定时刻刷新的周限额。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: Claude Code 里运行 /usage，或 claude.ai 的 Settings > Usage 看到它该怎么办 1. 换算竖线后的数字: 它是自 1970-01-01 UTC 起的秒数。贴到任意 epoch 转换工具，或在 macOS 上运行 date -r 1755615600，得到本地的刷新时间。 2. 运行 /usage: 它显示会话与周两条进度条和刷新时间，可以确认时间戳对应哪个窗口。 3. 等到时间再试: 时间戳之前的请求都会被拦。2025 年的版本不会自动恢复，时间到了要自己发下一条。 4. 更新 Claude Code: 现在的版本会用文字打印刷新时间，并且从 v2.1.234 起订阅登录的交互式会话可以原地等待、自动继续。 5. 用量正常仍反复出现就报 issue: 如果 /usage 显示还有余量但错误持续，带上文案和时间戳到 anthropics/claude-code 仓库提 issue。 什么时候重置？ 你自己的窗口：时间戳就是你窗口刷新的准确时刻。Anthropic 写明会话限额每 5 小时刷新、周限额在分配给你账号的固定时刻每周刷新。可以用下方链接的窗口倒计时记下换算后的时间。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: 竖线后的数字是什么意思？ A: 是 Unix 时间戳：自 1970-01-01 00:00 UTC 起的秒数。issue #6119 里的 1755615600 换算后是 2025-08-19 15:00 UTC，也就是那个账号窗口刷新的时刻。换算成本地时间就知道要等多久。 Q: 这句背后是什么错误？ A: API 的 429 rate_limit_error。issue #6119 的 errors 数组显示「This request would exceed your account's rate limit. Please try again later.」，Claude Platform 的错误码文档把 429 rate_limit_error 列为触发速率限制的代码。 Q: 这是 5 小时窗口还是周窗口？ A: 原始文案不说明。把时间戳和现在比：5 小时内是会话限额，几天后是周限额。/usage 直接显示两条进度条和各自的刷新时间。 Q: 为什么现在的版本看不到这句？ A: Error reference 现在记录的是格式化文案，例如「You've hit your session limit · resets 3:45pm」。竖线加时间戳的形态出现在 2025 年、v1.0.84 之类版本的报告里。 Sources: 1. anthropics/claude-code issue #6119 (2025-08-19): Claude AI usage limit reached|1755615600 (v1.0.84) — https://github.com/anthropics/claude-code/issues/6119 (verified 2026-09-17) 2. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) 3. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 4. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) ### "API Error: Server is temporarily limiting requests (not your usage limit)" — Claude Code URL: https://didreset.com/errors/claude-code-server-is-temporarily-limiting-requests-not-your-usage-limit Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: API Error: Server is temporarily limiting requests (not your usage limit) "API Error: Server is temporarily limiting requests (not your usage limit)" means the API applied a short-lived throttle that is unrelated to your plan quota; the message says so in brackets. There is no window to wait for: retry shortly and check status.claude.com if it persists. Since v2.1.199 Claude Code retries it automatically with backoff before showing it. Your session and weekly limits are untouched, and it is not an account-wide reset event. What does "API Error: Server is temporarily limiting requests (not your usage limit)" mean? The Error reference groups this message under usage limits but explains that it works differently: the API applied a short-lived throttle unrelated to your plan quota. Claude Code tells it apart from a real limit by the absence of the unified quota headers that a genuine limit response carries. As of v2.1.199 Claude Code retries it automatically with backoff before showing it, whichever way you authenticate. On earlier versions a session signed in with a claude.ai subscription failed the turn on the first occurrence; only API-key and Enterprise sign-ins retried it. Issue #68502 (2026-06-15) documents an HTTP 529 overloaded_error being rendered with this text and a "Rate limited" suffix, which made parallel sessions and subagents fail and led users to think they had hit a quota. The Claude Platform error reference lists 529 as an overloaded_error, separate from 429 rate_limit_error. Which window is full? - Window: None: a server-side throttle, not a usage window · Resets: No reset time; clears as soon as the server stops throttling · Where to check: status.claude.com for incidents; /usage to confirm your own windows still have room What to do when you see it 1. Wait briefly and try again: That is the docs' first instruction. Since v2.1.199 Claude Code has already retried with backoff before you see the message. 2. Check status.claude.com: The Error reference points there if the message persists; an incident on the status page explains a run of these errors. 3. Do not switch models or wait for a reset: The throttle is not tied to your session or weekly window, so neither /model nor waiting for your reset time changes anything. 4. Confirm your windows with /usage: If you want to be sure it is not a quota, /usage shows the session and weekly bars; a real limit would show "You've hit your session limit" instead. When does it reset? Your own window: nothing here is about your window. The session and weekly limits keep their normal schedule (every five hours, and at a fixed weekly time), and this error does not consume or extend them. If you want to know where you stand, /usage shows both bars. Account-wide resets: a throttle is also not an official reset. The board above records resets Anthropic announces, each linked to its original post. A run of these errors is more likely an incident on status.claude.com than anything the board would show. FAQ: Q: Is this my usage limit? A: No. The message itself says "not your usage limit", and the Error reference explains that the API applied a short-lived throttle unrelated to your plan quota. Claude Code distinguishes it from a real limit by the missing unified quota headers. Q: Do I need to wait for a reset? A: No. There is no window behind this error. Wait briefly and retry; since v2.1.199 Claude Code retries with backoff on its own before showing the message, whichever way you authenticate. Q: Why did it say "Rate limited" and break my subagents? A: Issue #68502 reports an HTTP 529 overloaded_error rendered with this text plus a "Rate limited" suffix, hard-failing parallel sessions and subagents on that version. The Platform error reference lists 529 as overloaded_error, distinct from the 429 rate_limit_error a real quota returns. Q: How can I tell this from a real limit? A: A real limit shows a message that names the window, such as "You've hit your session limit · resets 3:45pm", and /usage shows a bar at 100%. This message has no window and no reset time, and status.claude.com may show an incident. Sources: 1. Claude Code docs: Error reference — "Server is temporarily limiting requests" — https://code.claude.com/docs/en/errors#server-is-temporarily-limiting-requests (verified 2026-09-18) 2. anthropics/claude-code issue #68502 (2026-06-15): [Bug] HTTP 529 overloaded_error is rendered as "Rate limited" and hard-fails parallel sessions/subagents — https://github.com/anthropics/claude-code/issues/68502 (verified 2026-09-17) 3. Anthropic status page (status.claude.com) — https://status.claude.com (verified 2026-09-18) 4. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) ### 「API Error: Server is temporarily limiting requests (not your usage limit)」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-server-is-temporarily-limiting-requests-not-your-usage-limit Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: API Error: Server is temporarily limiting requests (not your usage limit) 「API Error: Server is temporarily limiting requests (not your usage limit)」表示 API 做了与套餐额度无关的短时节流。没有要等的窗口：稍后重试，持续出现看 status.claude.com。v2.1.199 起会先自动退避重试再显示。会话与周限额不受影响。 「API Error: Server is temporarily limiting requests (not your usage limit)」是什么意思？ Error reference 把这句放在用量限额一节，但说明它的性质不同：API 做了与套餐额度无关的短时节流。Claude Code 靠响应里缺少真正限额响应才会带的统一额度头来区分它。 v2.1.199 起，无论用哪种方式登录，Claude Code 都会先自动退避重试再显示它。更早的版本里，用 claude.ai 订阅登录的会话第一次遇到就会让本轮失败，只有 API key 与 Enterprise 登录会重试。 issue #68502（2026-06-15）记录了 HTTP 529 overloaded_error 被渲染成这句加「Rate limited」后缀，导致并行会话与子代理失败，也让用户误以为撞了额度。Claude Platform 的错误码文档把 529 列为 overloaded_error，与 429 rate_limit_error 不同。 满的是哪个窗口？ - 窗口: 没有：这是服务端节流，不是用量窗口 · 何时刷新: 没有刷新时间；服务端停止节流就恢复 · 在哪里看: status.claude.com 看事故；/usage 确认自己的窗口还有余量 看到它该怎么办 1. 稍等再试: 这是文档的第一条建议。v2.1.199 起，你看到这句之前 Claude Code 已经退避重试过了。 2. 看 status.claude.com: Error reference 指出持续出现就去看状态页；状态页上的事故能解释连续出现的这类错误。 3. 不要换模型，也不要等重置: 节流与你的会话或周窗口无关，/model 和等刷新时间都改变不了什么。 4. 用 /usage 确认窗口: 想确认不是额度问题，/usage 会显示会话与周进度条；真正的限额会显示「You've hit your session limit」。 什么时候重置？ 你自己的窗口：这句和你的窗口无关。会话与周限额照常按自己的节奏刷新（每 5 小时、每周固定时刻），这个错误既不消耗也不延长它们。想知道自己的状态，/usage 两条进度条都在。 全员重置：节流也不是官方重置。上方板面记录的是 Anthropic 宣布的重置，每条链接原帖。连续出现这类错误更可能对应 status.claude.com 上的事故，而不是板面会显示的任何事件。 FAQ: Q: 这是我的额度用完了吗？ A: 不是。文案本身写着「not your usage limit」，Error reference 也说明 API 做的是与套餐额度无关的短时节流。Claude Code 靠缺少统一额度头把它和真正的限额区分开。 Q: 需要等重置吗？ A: 不需要。这个错误背后没有窗口。稍等再试即可；v2.1.199 起，无论哪种登录方式，Claude Code 都会先自动退避重试再显示这句。 Q: 为什么它显示「Rate limited」还弄挂了我的子代理？ A: issue #68502 报告 HTTP 529 overloaded_error 在那个版本上被渲染成这句加「Rate limited」后缀，并让并行会话与子代理直接失败。Platform 错误码文档把 529 列为 overloaded_error，与真正额度用尽返回的 429 rate_limit_error 不同。 Q: 怎么和真正的限额区分？ A: 真正的限额会显示写明窗口的文案，例如「You've hit your session limit · resets 3:45pm」，/usage 里有一条进度条到 100%。这句没有窗口也没有刷新时间，status.claude.com 上可能有事故记录。 Sources: 1. Claude Code docs: Error reference — "Server is temporarily limiting requests" — https://code.claude.com/docs/en/errors#server-is-temporarily-limiting-requests (verified 2026-09-18) 2. anthropics/claude-code issue #68502 (2026-06-15): [Bug] HTTP 529 overloaded_error is rendered as "Rate limited" and hard-fails parallel sessions/subagents — https://github.com/anthropics/claude-code/issues/68502 (verified 2026-09-17) 3. Anthropic status page (status.claude.com) — https://status.claude.com (verified 2026-09-18) 4. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) ### "API Error: Usage credits required for 1M context · run /usage-credits to turn them on (they take effect after you restart Claude Code), or /model to switch to standard context" — Claude Code URL: https://didreset.com/errors/claude-code-usage-credits-required-for-1m-context Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: API Error: Usage credits required for 1M context · run /usage-credits to turn them on (they take effect after you restart Claude Code), or /model to switch to standard context "API Error: Usage credits required for 1M context · run /usage-credits to turn them on (they take effect after you restart Claude Code), or /model to switch to standard context" means the model you picked uses the 1M-token context window and your plan includes it only through usage credits. The Error reference calls it an entitlement check, not a quota exhaustion: it fires even with session and weekly capacity remaining. No reset clears it; /model or /usage-credits does. What does "API Error: Usage credits required for 1M context · run /usage-credits to turn them on (they take effect after you restart Claude Code), or /model to switch to standard context" mean? The selected model uses the 1M-token extended context window, and your plan only includes it through usage credits. The Error reference is explicit: "This is an entitlement check, not a quota exhaustion. It fires even when your session and weekly allowances have capacity remaining." Claude Code runs the check when you pick the model with /model, and only on a direct connection to the Anthropic API. When the error appears mid-conversation because the context grew past 200K tokens, current versions compact the conversation back under the standard limit automatically; before v2.1.172 the error repeated on every request and /clear was the way out. The wording has changed. Before v2.1.268 the message ended with "run /usage-credits to turn them on, or /model to switch to standard context" and did not mention restarting. An even older variant, reported in issue #61828 in May 2026, told users to turn on usage credits at claude.ai/settings/usage or use --model; that issue was filed as "Usage limit reached" with session usage at 2% and weekly at 32%. Which window is full? - Window: None: an entitlement check, not a usage window · Resets: Never; the error does not depend on how much you have used · Where to check: /usage shows the session and weekly bars, which will still have room What to do when you see it 1. Run /model and pick the variant without [1m]: That falls back to the standard context window. It is the docs' first step. 2. Or turn on usage credits: Where the message names /usage-credits, run it to turn on metered billing for the 1M variant on Pro and Max, or to request usage credits from your admin on Team and Enterprise. Restart Claude Code afterwards; until you do, the session stays at the standard limit. 3. Check where the 1M model is set: If the error persists after /model, a 1M model ID may be set in a configuration location that takes precedence; the docs list them under Setting your model. 4. Remove 1M variants from the picker: Set CLAUDE_CODE_DISABLE_1M_CONTEXT=1 to take the [1m] models out of /model entirely. When does it reset? Your own window: this error is not caused by a full window, so no reset clears it. Your session limit keeps resetting every five hours and your weekly limit at its fixed time, and the help center notes that usage credits do not affect that timing; the fix is a model choice or usage credits, not waiting. Account-wide resets: an official reset does not change plan entitlements either. The board above records resets Anthropic announces, each linked to its original post; none of them turns 1M context on for a plan that includes it only through usage credits. FAQ: Q: Will waiting for my session or weekly reset fix this? A: No. The Error reference says it is an entitlement check, not a quota exhaustion, and that it fires even when your session and weekly allowances have capacity remaining. Switch to a standard-context model with /model or turn on usage credits. Q: Why does it say usage limit reached when /usage shows 2%? A: Because an older wording was easy to misread. Issue #61828 (2026-05-23) reports "Usage limit reached" with session usage at 2% and weekly at 32%; the message in that report was the older 1M-context variant pointing to claude.ai/settings/usage and --model. Current builds say "Usage credits required for 1M context". Q: What changed in v2.1.268? A: The message now says the credits take effect after you restart Claude Code. Before v2.1.268 it ended with "run /usage-credits to turn them on, or /model to switch to standard context" and did not mention restarting. Q: It appeared in the middle of a long session. Do I need to do anything? A: Usually not. When the context grows past 200K tokens, current versions compact the conversation back under the standard limit and keep the session there. Before v2.1.172 the error repeated on every request, including /compact, and /clear was the recovery. Sources: 1. Claude Code docs: Error reference — "Usage credits required for 1M context" — https://code.claude.com/docs/en/errors#usage-credits-required-for-1m-context (verified 2026-09-18) 2. anthropics/claude-code issue #61828 (2026-05-23): [BUG] Claude Code shows 'Usage limit reached' despite session at 2% and weekly at 32% (older 1M-context wording) — https://github.com/anthropics/claude-code/issues/61828 (verified 2026-09-17) 3. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 4. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) ### 「API Error: Usage credits required for 1M context · run /usage-credits to turn them on (they take effect after you restart Claude Code), or /model to switch to standard context」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-usage-credits-required-for-1m-context Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: API Error: Usage credits required for 1M context · run /usage-credits to turn them on (they take effect after you restart Claude Code), or /model to switch to standard context 「API Error: Usage credits required for 1M context …」表示你选的模型使用 1M token 上下文，而你的套餐只能通过 usage credits 使用它。Error reference 称之为权益检查而非额度用尽：额度有余量也会触发。等重置没用，/model 或 /usage-credits 才有用。 「API Error: Usage credits required for 1M context · run /usage-credits to turn them on (they take effect after you restart Claude Code), or /model to switch to standard context」是什么意思？ 所选模型使用 1M token 的扩展上下文，而你的套餐只能通过 usage credits 使用它。Error reference 写得很清楚：「This is an entitlement check, not a quota exhaustion. It fires even when your session and weekly allowances have capacity remaining.」 Claude Code 在你用 /model 选模型时做这项检查，且只在直连 Anthropic API 时做。如果是对话中途上下文超过 200K token 才出现，现在的版本会自动把对话压缩回标准上限；v2.1.172 之前它会在每次请求时重复出现，只能用 /clear 恢复。 文案变过。v2.1.268 之前结尾是「run /usage-credits to turn them on, or /model to switch to standard context」，不提重启。issue #61828（2026 年 5 月）里还有更早的变体，让用户去 claude.ai/settings/usage 开通 usage credits 或用 --model；那条 issue 以「Usage limit reached」上报，而会话用量只有 2%、周用量 32%。 满的是哪个窗口？ - 窗口: 没有：这是权益检查，不是用量窗口 · 何时刷新: 不会刷新；这个错误与你用了多少无关 · 在哪里看: /usage 显示会话与周进度条，它们仍会有余量 看到它该怎么办 1. 运行 /model 选不带 [1m] 的模型: 这样回到标准上下文窗口，是文档的第一步。 2. 或者开通 usage credits: 文案里提到 /usage-credits 时，运行它在 Pro / Max 上为 1M 变体开通按量计费，或在 Team / Enterprise 上向管理员申请。之后重启 Claude Code；重启前会话仍停在标准上限。 3. 检查 1M 模型设在哪里: 如果 /model 之后仍报错，可能有更高优先级的配置位置设了 1M 模型 ID；文档在 Setting your model 一节列出了这些位置。 4. 把 1M 变体从选单里去掉: 设置 CLAUDE_CODE_DISABLE_1M_CONTEXT=1，/model 里就不再出现 [1m] 模型。 什么时候重置？ 你自己的窗口：这个错误不是窗口满了引起的，所以任何刷新都清不掉它。会话限额照常每 5 小时刷新、周限额在固定时刻刷新，帮助中心也说明 usage credits 不影响这个时间；解决办法是换模型或开通 usage credits，不是等。 全员重置：官方重置也不会改变套餐权益。上方板面记录的是 Anthropic 宣布的重置，每条链接原帖；没有哪一次会给只能靠 usage credits 用 1M 的套餐打开 1M 上下文。 FAQ: Q: 等会话或周限额刷新能解决吗？ A: 不能。Error reference 写明这是权益检查而非额度用尽，会话与周额度还有余量时也会触发。用 /model 切到标准上下文的模型，或开通 usage credits。 Q: 为什么 /usage 显示 2% 它却说 usage limit reached？ A: 因为旧文案容易误读。issue #61828（2026-05-23）以「Usage limit reached」上报，会话用量 2%、周用量 32%；那份报告里的文案是更早的 1M 上下文变体，指向 claude.ai/settings/usage 与 --model。现在的版本写的是「Usage credits required for 1M context」。 Q: v2.1.268 改了什么？ A: 现在的文案会说明 credits 在重启 Claude Code 后生效。v2.1.268 之前结尾是「run /usage-credits to turn them on, or /model to switch to standard context」，不提重启。 Q: 长会话中途出现，需要做什么吗？ A: 通常不需要。上下文超过 200K token 时，现在的版本会自动把对话压缩回标准上限并保持在那里。v2.1.172 之前它会在每次请求（包括 /compact）时重复出现，要用 /clear 恢复。 Sources: 1. Claude Code docs: Error reference — "Usage credits required for 1M context" — https://code.claude.com/docs/en/errors#usage-credits-required-for-1m-context (verified 2026-09-18) 2. anthropics/claude-code issue #61828 (2026-05-23): [BUG] Claude Code shows 'Usage limit reached' despite session at 2% and weekly at 32% (older 1M-context wording) — https://github.com/anthropics/claude-code/issues/61828 (verified 2026-09-17) 3. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) 4. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) ### "Not logged in · Please run /login" — Claude Code URL: https://didreset.com/errors/claude-code-not-logged-in-please-run-login Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Not logged in · Please run /login Variants: Login expired · Please run /login | Failed to authenticate: OAuth session expired and could not be refreshed "Not logged in · Please run /login" is an authentication error, not a usage limit: the Error reference says no valid credential is available for this session. It has nothing to do with your five-hour or weekly window, so waiting for a reset will not clear it. Run /login to sign in with your Claude subscription or Console account, or check that ANTHROPIC_API_KEY is exported in the shell where you launched claude. /status shows which credential is active. What does "Not logged in · Please run /login" mean? The Error reference files this under "Authentication errors": "These errors mean Claude Code cannot prove who you are to the API." For this message specifically, "No valid credential is available for this session." Nothing has been counted against a quota; the request never got that far. A close relative is "Login expired · Please run /login": Claude Code tried to renew your saved claude.ai or Console login, the OAuth service rejected the stored refresh token, and Claude Code cleared the saved credentials. In non-interactive mode the same state reads "Failed to authenticate: OAuth session expired and could not be refreshed". Either way, only /login creates new credentials. It is easy to mistake for a limit because it also stops every request. The difference: a usage limit names a window and a reset time ("You've hit your session limit · resets 3:45pm"); this message names none, because there is no window involved. This is not a quota problem, and a reset will not solve it. Which window is full? - Window: None: this is an authentication failure, not a usage window · Resets: Never; there is nothing to wait for · Where to check: /status shows which credential is active What to do when you see it 1. Run /login: Authenticate with your Claude subscription or Console account. The help center adds that if you were logged in via Console pay-as-you-go, /login switches you to your subscription plan. 2. Check the environment variable if you expected one: Confirm ANTHROPIC_API_KEY is set and exported in the shell where you launched claude. Tools that load a .env file can also leave a stale key. 3. For CI, use an apiKeyHelper: Where interactive login is not possible, the docs recommend an apiKeyHelper script that fetches a key at startup. 4. Read the precedence order: When several credentials are present, Claude Code picks one in a fixed order: cloud provider variables, ANTHROPIC_AUTH_TOKEN, ANTHROPIC_API_KEY, apiKeyHelper, CLAUDE_CODE_OAUTH_TOKEN, profiles, then the /login subscription. /status shows which one won. 5. If you keep getting logged out: The docs point to system clock checks and macOS credential-storage recovery under "Not logged in or token expired" in the install troubleshooting page. When does it reset? Your own window: nothing resets here. This error is about credentials, not capacity, so your five-hour and weekly windows are untouched and irrelevant. After /login succeeds, /usage shows where those windows stand. Account-wide resets: an official reset announced by Anthropic would not sign you in either. The board above records those announcements for people who are signed in and waiting on a quota; this page is not one of those cases. FAQ: Q: Is "Not logged in" a usage limit? A: No. The Error reference groups it under authentication errors and says no valid credential is available for this session. A usage limit would name a window and a reset time. This is not a quota problem, and waiting for any reset will not solve it. Q: What is the difference from "Login expired · Please run /login"? A: "Login expired" means Claude Code tried to renew a saved login, the OAuth service rejected the refresh token, and the saved credentials were cleared. "Not logged in" means no credential was available at all. Both are fixed the same way: run /login. Q: I set ANTHROPIC_API_KEY. Why am I still asked to log in? A: The docs say to confirm the key is set and exported in the shell where you launched claude. If a key is present but rejected, you would see "Invalid API key" instead. /status shows which credential source Claude Code actually resolved. Q: Will /login switch me from API pay-as-you-go to my subscription? A: Yes. Anthropic's help center says that if you are already logged in via Claude Console pay-as-you-go, running /login from within Claude Code switches you to your subscription plan. Sources: 1. Claude Code docs: Error reference — "Not logged in" — https://code.claude.com/docs/en/errors#not-logged-in (verified 2026-09-18) 2. Claude Code docs: Error reference — "Login expired" — https://code.claude.com/docs/en/errors#login-expired (verified 2026-09-18) 3. Claude Code docs: Authentication — Authentication precedence — https://code.claude.com/docs/en/authentication#authentication-precedence (verified 2026-09-18) 4. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) ### 「Not logged in · Please run /login」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-not-logged-in-please-run-login Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Not logged in · Please run /login Variants: Login expired · Please run /login | Failed to authenticate: OAuth session expired and could not be refreshed 「Not logged in · Please run /login」是认证错误，不是用量限额：Error reference 写明本次会话没有可用的有效凭证。它和 5 小时窗口、周窗口无关，等重置不会消除它。运行 /login 用 Claude 订阅或 Console 账号登录，或确认 shell 里导出了 ANTHROPIC_API_KEY。 「Not logged in · Please run /login」是什么意思？ Error reference 把它归在「Authentication errors」：「These errors mean Claude Code cannot prove who you are to the API.」这一句具体指「No valid credential is available for this session.」没有任何用量被计入额度，请求根本没走到那一步。 近亲是「Login expired · Please run /login」：Claude Code 尝试续期你保存的 claude.ai 或 Console 登录，OAuth 服务拒绝了存储的 refresh token，Claude Code 于是清除了保存的凭证。非交互模式下同一状态显示「Failed to authenticate: OAuth session expired and could not be refreshed」。无论哪种，只有 /login 能生成新凭证。 它容易被当成限额，因为同样会拦下所有请求。区别在于：用量限额会写明窗口和刷新时间（「You've hit your session limit · resets 3:45pm」），这句什么都不写，因为根本没有窗口。这不是额度问题，重置也不会解决。 满的是哪个窗口？ - 窗口: 没有：这是认证失败，不是额度窗口 · 何时刷新: 不会刷新；没有可等的时间 · 在哪里看: /status 显示当前生效的凭证 看到它该怎么办 1. 运行 /login: 用 Claude 订阅或 Console 账号登录。帮助中心补充：如果你之前用 Console 按量付费登录，/login 会切换到你的订阅计划。 2. 期望用环境变量就检查它: 确认 ANTHROPIC_API_KEY 在启动 claude 的那个 shell 里已设置并导出。会加载 .env 的工具也可能留下失效的 key。 3. CI 场景用 apiKeyHelper: 无法交互登录时，文档建议配置一个在启动时获取 key 的 apiKeyHelper 脚本。 4. 了解凭证优先级: 同时存在多个凭证时，Claude Code 按固定顺序选择：云厂商变量、ANTHROPIC_AUTH_TOKEN、ANTHROPIC_API_KEY、apiKeyHelper、CLAUDE_CODE_OAUTH_TOKEN、profile，最后才是 /login 的订阅登录。/status 显示最终选中的是哪个。 5. 反复被登出: 文档指向安装故障排查页的「Not logged in or token expired」一节：检查系统时钟、修复 macOS 凭证存储。 什么时候重置？ 你自己的窗口：这里没有任何东西会刷新。这个错误关于凭证而不是额度，所以 5 小时窗口与周窗口既没受影响也无关。/login 成功后，/usage 会显示那些窗口的状态。 全员重置：Anthropic 宣布的官方重置也不会替你登录。上方板面记录那些公告，是给已登录、正在等额度的人看的；这一页不属于那种情况。 FAQ: Q: 「Not logged in」是用量限额吗？ A: 不是。Error reference 把它归在认证错误，说明本次会话没有可用的有效凭证。用量限额会写明窗口和刷新时间。这不是额度问题，等任何重置都不会解决。 Q: 它和「Login expired · Please run /login」有什么区别？ A: 「Login expired」表示 Claude Code 尝试续期保存的登录时，OAuth 服务拒绝了 refresh token，保存的凭证被清除。「Not logged in」表示根本没有可用凭证。两者的修法一样：运行 /login。 Q: 我设了 ANTHROPIC_API_KEY，为什么还让我登录？ A: 文档要求确认 key 在启动 claude 的那个 shell 里已设置并导出。如果 key 存在但被拒绝，你看到的会是「Invalid API key」。/status 显示 Claude Code 实际解析到的凭证来源。 Q: /login 能把我从 API 按量付费切到订阅吗？ A: 能。Anthropic 帮助中心写明，如果你已经用 Claude Console 按量付费登录，在 Claude Code 里运行 /login 会切换到你的订阅计划。 Sources: 1. Claude Code docs: Error reference — "Not logged in" — https://code.claude.com/docs/en/errors#not-logged-in (verified 2026-09-18) 2. Claude Code docs: Error reference — "Login expired" — https://code.claude.com/docs/en/errors#login-expired (verified 2026-09-18) 3. Claude Code docs: Authentication — Authentication precedence — https://code.claude.com/docs/en/authentication#authentication-precedence (verified 2026-09-18) 4. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) ### "Invalid API key · Fix external API key" — Claude Code URL: https://didreset.com/errors/claude-code-invalid-api-key-fix-external-api-key Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Invalid API key · Fix external API key "Invalid API key · Fix external API key" means the ANTHROPIC_API_KEY environment variable or an apiKeyHelper script returned a key the API rejected, or Claude Code blocked a malformed key before sending it. It is an authentication error, not a usage limit: no window is full and no reset will clear it. Check the key in the Claude Console, run env | grep ANTHROPIC to find a stale key, or unset the variable and run /login to use your subscription. What does "Invalid API key · Fix external API key" mean? The Error reference lists this under authentication errors: "The ANTHROPIC_API_KEY environment variable or apiKeyHelper script returned a key the API rejected, or Claude Code blocked a key from ANTHROPIC_API_KEY before sending it." At the API level a rejected key is a 401 authentication_error, which the Platform docs describe as an issue with your API key, for example malformed, revoked or expired. The word "external" matters: the key came from your environment, not from /login. Environment variables take precedence over a saved login, so a key exported in your shell profile or loaded from a .env file is used even when you have a working Pro or Max subscription. The docs note that direnv, dotenv shell plugins and IDE terminals can load a stale key without you setting it explicitly. When the message continues with a description such as "it contains a line break at character 41", the API never saw the key: Claude Code found a character HTTP headers cannot carry and stopped the request. Either way, this is not a quota problem, and a reset will not solve it. Which window is full? - Window: None: this is an authentication failure, not a usage window · Resets: Never; there is nothing to wait for · Where to check: /status shows which credential is active What to do when you see it 1. Check the key in the Console: Look for typos and confirm the key has not been revoked at platform.claude.com/settings/keys. 2. Find where the key comes from: In the same shell run env | grep ANTHROPIC (PowerShell: Get-ChildItem Env:ANTHROPIC*). A .env file in the project, direnv or an IDE terminal may be supplying a stale key. 3. Or switch to your subscription: Unset ANTHROPIC_API_KEY and run /login to use subscription auth instead. The help center says /login switches you from Console pay-as-you-go to your subscription plan. 4. If an apiKeyHelper supplies the key: Run the script directly to confirm it prints a valid key on stdout, as the docs suggest. 5. Confirm with /status: It shows which credential source Claude Code is actually using, so you can see whether the fix took. When does it reset? Your own window: nothing resets here. The request was rejected before any usage was counted, so your five-hour and weekly windows are untouched. Once a valid credential is in place, /usage shows where they stand. Account-wide resets: an official reset announced by Anthropic changes quotas, not credentials. The board above records those announcements; a bad API key stays bad through any reset. FAQ: Q: Is "Invalid API key" a usage limit? A: No. The Error reference lists it under authentication errors, and the underlying API code is 401 authentication_error, which the Platform docs describe as an issue with your API key. No window is involved; waiting for a reset changes nothing. Q: I have a Max subscription. Why is an API key being used at all? A: Because environment variables take precedence over /login. If ANTHROPIC_API_KEY is exported in your shell profile or loaded from a .env file, Claude Code uses it once approved. Unset it and run /login to route through your subscription. Q: What does the text after "Fix external API key" mean? A: When it continues with a description such as a line break at a given character position, Claude Code found a character HTTP headers cannot carry and stopped the request before sending it. The API never saw the key; fix the value in the variable or file it came from. Q: How do I know which credential Claude Code picked? A: Run /status: it shows which credential source Claude Code is actually using. The authentication docs describe a fixed precedence when several are present: cloud provider variables first, then ANTHROPIC_AUTH_TOKEN, ANTHROPIC_API_KEY, apiKeyHelper, CLAUDE_CODE_OAUTH_TOKEN, profiles, and last the /login subscription. Sources: 1. Claude Code docs: Error reference — "Invalid API key" — https://code.claude.com/docs/en/errors#invalid-api-key (verified 2026-09-18) 2. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) 3. Claude Code docs: Authentication — Authentication precedence — https://code.claude.com/docs/en/authentication#authentication-precedence (verified 2026-09-18) 4. Claude Console: API keys — https://platform.claude.com/settings/keys (verified 2026-09-18) 5. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) ### 「Invalid API key · Fix external API key」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-invalid-api-key-fix-external-api-key Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Invalid API key · Fix external API key 「Invalid API key · Fix external API key」表示 ANTHROPIC_API_KEY 或 apiKeyHelper 给出的 key 被 API 拒绝，或格式不对被拦下。它是认证错误，不是用量限额：没有窗口满了，任何重置都清不掉。到 Claude Console 检查 key，或取消该变量并运行 /login 改用订阅。 「Invalid API key · Fix external API key」是什么意思？ Error reference 把它列在认证错误里：「The ANTHROPIC_API_KEY environment variable or apiKeyHelper script returned a key the API rejected, or Claude Code blocked a key from ANTHROPIC_API_KEY before sending it.」在 API 层面，被拒绝的 key 对应 401 authentication_error，Platform 文档描述为 API key 有问题，例如格式错误、已撤销或已过期。 「external」这个词很关键：key 来自你的环境，而不是 /login。环境变量优先于保存的登录，所以写在 shell 配置或 .env 文件里的 key 即使在 Pro / Max 订阅有效时也会被使用。文档提醒 direnv、dotenv 类 shell 插件和 IDE 终端都可能在你没有显式设置的情况下加载失效的 key。 如果文案后面跟着「it contains a line break at character 41」之类的描述，API 根本没看到这个 key：Claude Code 发现了 HTTP 头无法承载的字符，在发送前就拦下了请求。无论哪种情况，这不是额度问题，重置也不会解决。 满的是哪个窗口？ - 窗口: 没有：这是认证失败，不是额度窗口 · 何时刷新: 不会刷新；没有可等的时间 · 在哪里看: /status 显示当前生效的凭证 看到它该怎么办 1. 到 Console 检查 key: 在 platform.claude.com/settings/keys 检查是否有拼写错误、是否已被撤销。 2. 找出 key 从哪来: 在同一个 shell 里运行 env | grep ANTHROPIC（PowerShell：Get-ChildItem Env:ANTHROPIC*）。项目里的 .env、direnv 或 IDE 终端都可能在提供失效的 key。 3. 或者改用订阅: 取消 ANTHROPIC_API_KEY 并运行 /login 改用订阅认证。帮助中心说明 /login 会把你从 Console 按量付费切到订阅计划。 4. key 来自 apiKeyHelper 时: 按文档建议直接运行那个脚本，确认它在 stdout 打印出有效的 key。 5. 用 /status 确认: 它显示 Claude Code 实际使用的凭证来源，可以看到修复是否生效。 什么时候重置？ 你自己的窗口：这里没有任何东西会刷新。请求在计入任何用量之前就被拒绝了，5 小时窗口与周窗口都没动。凭证正确后，/usage 会显示它们的状态。 全员重置：Anthropic 宣布的官方重置改变的是额度，不是凭证。上方板面记录那些公告；坏掉的 API key 经过任何重置都还是坏的。 FAQ: Q: 「Invalid API key」是用量限额吗？ A: 不是。Error reference 把它列在认证错误里，底层 API 错误码是 401 authentication_error，Platform 文档描述为 API key 有问题。这里不涉及任何窗口，等重置什么都不会改变。 Q: 我有 Max 订阅，为什么会用到 API key？ A: 因为环境变量优先于 /login。如果 ANTHROPIC_API_KEY 写在 shell 配置里或从 .env 文件加载，Claude Code 在你批准后就会使用它。取消该变量并运行 /login 才会走订阅。 Q: 「Fix external API key」后面的文字是什么意思？ A: 如果后面跟着某个字符位置有换行之类的描述，说明 Claude Code 发现了 HTTP 头无法承载的字符，在发送前就拦下了请求。API 根本没看到这个 key；去它来源的变量或文件里修正值。 Q: 怎么知道 Claude Code 选了哪个凭证？ A: 运行 /status。文档描述了固定的优先级：云厂商变量、ANTHROPIC_AUTH_TOKEN、ANTHROPIC_API_KEY、apiKeyHelper、CLAUDE_CODE_OAUTH_TOKEN、profile，最后才是 /login 的订阅登录。 Sources: 1. Claude Code docs: Error reference — "Invalid API key" — https://code.claude.com/docs/en/errors#invalid-api-key (verified 2026-09-18) 2. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) 3. Claude Code docs: Authentication — Authentication precedence — https://code.claude.com/docs/en/authentication#authentication-precedence (verified 2026-09-18) 4. Claude Console: API keys — https://platform.claude.com/settings/keys (verified 2026-09-18) 5. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) ### "Please run /login · API Error: 401 Invalid authentication credentials" — Claude Code URL: https://didreset.com/errors/claude-code-please-run-login-api-error-401-invalid-authentication-credentials Language: en · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Please run /login · API Error: 401 Invalid authentication credentials "Please run /login · API Error: 401 Invalid authentication credentials" is an authentication failure, not a usage limit. The Error reference says the API recognized the format of your credential but rejected the account or organization behind it: a revoked credential, a disabled organization, or a deactivated account. At the API level it is a 401 authentication_error. No window is full and no reset will clear it; run /status to see the active credential, then rotate the key or /login. What does "Please run /login · API Error: 401 Invalid authentication credentials" mean? The Error reference describes this as the API recognizing the format of your credential but rejecting the account or organization behind it. Anthropic returns it when a credential was recently revoked, when an organization was disabled or removed your access, or when the account itself was deactivated; the docs add that an expired token is not the cause. The HTTP status is 401. The Claude Platform error reference lists 401 as authentication_error, "an issue with your API key (for example, it's malformed, revoked, or expired)". A usage limit is a different code entirely: 429 rate_limit_error. That is the whole distinction: this is not a quota problem, and a reset will not solve it. The credential can be your saved /login or an approved ANTHROPIC_API_KEY, and the fix differs. Environment variables take precedence over /login, so if /status shows an API key row that is in use, running /login does not replace it. If ANTHROPIC_BASE_URL points at an LLM gateway, the text after 401 is the gateway's message rather than Anthropic's. Which window is full? - Window: None: this is an authentication failure, not a usage window · Resets: Never; there is nothing to wait for · Where to check: /status shows which credential is active What to do when you see it 1. Run /status: See which credential is active: an API key row that is not marked as not in use means an approved ANTHROPIC_API_KEY is winning over your login. 2. If an API key is active: Rotate the key in the Claude Console, or fall back to your subscription with unset ANTHROPIC_API_KEY (PowerShell: Remove-Item Env:ANTHROPIC_API_KEY). 3. If only your login is active: Run /login once. If the credential was revoked, a fresh login replaces it. The help center notes /login also switches you from Console pay-as-you-go to your subscription plan. 4. If the same message returns for the same account: The account or organization is no longer active. Check the account and organization that /status reports and ask your organization admin to restore access. 5. If you use an LLM gateway: With ANTHROPIC_BASE_URL set, the text after 401 comes from the gateway; fix the credential the gateway expects instead of running /login. When does it reset? Your own window: nothing resets here. A 401 is returned before any usage is counted, so your five-hour and weekly windows are untouched; when a usage window is actually full, Claude Code shows "You've hit your session limit" with a reset time instead. After access is restored, /usage shows where the windows stand. Account-wide resets: an official reset announced by Anthropic restores quota for accounts in good standing; it does not reactivate a revoked credential or a disabled organization. The board above records those announcements for a different situation than this one. FAQ: Q: Is a 401 a usage limit? A: No. The Platform error reference lists 401 as authentication_error, an issue with your API key or credential. A usage limit returns 429 rate_limit_error, and Claude Code shows it as "You've hit your session limit" or a sibling message with a reset time. Waiting for a reset does not change a 401. Q: Why does /login not fix it? A: Usually because an approved ANTHROPIC_API_KEY is the active credential. Environment variables take precedence over your login, so /login does not replace the key. Rotate it in the Console or unset the variable. If /status shows only your login and the message persists, the account or organization is no longer active. Q: Is my token just expired? A: The Error reference says an expired token is not the cause of this message: the API rejected the account or organization behind a well-formed credential, after a revocation, a disabled organization, or a deactivated account. An expired saved login shows "Login expired · Please run /login" instead. Q: How is this different from "Invalid API key · Fix external API key"? A: "Invalid API key" means the key itself was rejected or malformed. This 401 message means the credential's format was fine but the account or organization behind it was rejected. Both are authentication errors with no window and no reset time. Sources: 1. Claude Code docs: Error reference — "API Error: 401 Invalid authentication credentials" — https://code.claude.com/docs/en/errors#api-error-401-invalid-authentication-credentials (verified 2026-09-18) 2. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) 3. Claude Code docs: Authentication — Authentication precedence — https://code.claude.com/docs/en/authentication#authentication-precedence (verified 2026-09-18) 4. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 5. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) ### 「Please run /login · API Error: 401 Invalid authentication credentials」 — Claude Code URL: https://didreset.com/zh/errors/claude-code-please-run-login-api-error-401-invalid-authentication-credentials Language: zh · Surface: CLI · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Please run /login · API Error: 401 Invalid authentication credentials 「Please run /login · API Error: 401 Invalid authentication credentials」是认证失败，不是用量限额。Error reference 写明 API 认得凭证格式，但拒绝了背后的账号或组织：凭证被撤销、组织被停用或账号被注销。重置清不掉它；先 /status，再 /login。 「Please run /login · API Error: 401 Invalid authentication credentials」是什么意思？ Error reference 描述它为：API 认得你凭证的格式，却拒绝了背后的账号或组织。Anthropic 在凭证刚被撤销、组织被停用或移除了你的访问权限、或账号本身被注销时返回它；文档还补充，token 过期不是原因。 HTTP 状态码是 401。Claude Platform 错误码文档把 401 列为 authentication_error，「an issue with your API key (for example, it's malformed, revoked, or expired)」。用量限额是另一个码：429 rate_limit_error。区别就在这里：这不是额度问题，重置也不会解决。 凭证可能是你保存的 /login，也可能是已批准的 ANTHROPIC_API_KEY，修法不同。环境变量优先于 /login，所以如果 /status 显示 API key 那一行在使用中，运行 /login 并不会替换它。如果 ANTHROPIC_BASE_URL 指向 LLM 网关，401 后面的文字是网关的而不是 Anthropic 的。 满的是哪个窗口？ - 窗口: 没有：这是认证失败，不是额度窗口 · 何时刷新: 不会刷新；没有可等的时间 · 在哪里看: /status 显示当前生效的凭证 看到它该怎么办 1. 运行 /status: 看当前生效的是哪个凭证：API key 那一行没有标为未使用，就说明已批准的 ANTHROPIC_API_KEY 压过了你的登录。 2. 如果生效的是 API key: 到 Claude Console 轮换 key，或者用 unset ANTHROPIC_API_KEY（PowerShell：Remove-Item Env:ANTHROPIC_API_KEY）退回订阅。 3. 如果只有登录在生效: 运行一次 /login。若凭证被撤销，新的登录会替换它。帮助中心也说明 /login 会把你从 Console 按量付费切到订阅计划。 4. 同一账号反复出现同一句: 说明账号或组织已不再有效。核对 /status 报告的账号与组织，请组织管理员恢复访问权限。 5. 如果你用 LLM 网关: 设置了 ANTHROPIC_BASE_URL 时，401 后面的文字来自网关；去修网关要求的凭证，而不是运行 /login。 什么时候重置？ 你自己的窗口：这里没有任何东西会刷新。401 在计入任何用量之前就返回了，5 小时窗口与周窗口都没动；窗口真的满了时，Claude Code 显示的是带刷新时间的「You've hit your session limit」。访问恢复后，/usage 会显示窗口状态。 全员重置：Anthropic 宣布的官方重置给状态正常的账号恢复额度，不会重新激活被撤销的凭证或被停用的组织。上方板面记录那些公告，对应的是和这一页不同的情况。 FAQ: Q: 401 是用量限额吗？ A: 不是。Platform 错误码文档把 401 列为 authentication_error，即 API key 或凭证的问题。用量限额返回的是 429 rate_limit_error，Claude Code 会显示成带刷新时间的「You've hit your session limit」或同类文案。等重置改变不了 401。 Q: 为什么 /login 没用？ A: 通常因为已批准的 ANTHROPIC_API_KEY 才是生效的凭证。环境变量优先于登录，所以 /login 替换不了这个 key。到 Console 轮换它或取消该变量。如果 /status 只显示登录而错误依旧，说明账号或组织已不再有效。 Q: 是不是我的 token 过期了？ A: Error reference 写明 token 过期不是这句的原因：API 拒绝的是格式正确的凭证背后的账号或组织，起因是撤销、组织停用或账号注销。保存的登录过期显示的是「Login expired · Please run /login」。 Q: 它和「Invalid API key · Fix external API key」有什么区别？ A: 「Invalid API key」表示 key 本身被拒绝或格式不对。这句 401 表示凭证格式没问题，但背后的账号或组织被拒绝。两者都是认证错误，没有窗口，也没有刷新时间。 Sources: 1. Claude Code docs: Error reference — "API Error: 401 Invalid authentication credentials" — https://code.claude.com/docs/en/errors#api-error-401-invalid-authentication-credentials (verified 2026-09-18) 2. Claude Platform docs: API errors (401 authentication_error, 429 rate_limit_error) — https://platform.claude.com/docs/en/api/errors (verified 2026-09-18) 3. Claude Code docs: Authentication — Authentication precedence — https://code.claude.com/docs/en/authentication#authentication-precedence (verified 2026-09-18) 4. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 5. Claude Code docs: Error reference — "You've hit your session limit" — https://code.claude.com/docs/en/errors#youve-hit-your-session-limit (verified 2026-09-18) ### "5-hour limit reached - resets {time}." — Claude URL: https://didreset.com/errors/claude-5-hour-limit-reached-resets Language: en · Surface: Web · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: 5-hour limit reached - resets {time}. Variants: Approaching 5-hour limit. | 5-hour limit resets [time] - continuing with usage credits. "5-hour limit reached - resets {time}." is the blocking message in the Claude web and desktop apps when your five-hour session allowance is used up; the help center lists it after the warning "Approaching 5-hour limit." Anthropic says the session limit resets every five hours, and the time in the message is your own clock: Settings > Usage shows it. With usage credits on, the message changes to "continuing with usage credits". It is not an account-wide reset. What does "5-hour limit reached - resets {time}." mean? Anthropic's "Troubleshoot Claude error messages" article describes three usage-limit messages in the Claude apps. As you approach the limit within a five-hour session you see "Approaching 5-hour limit."; if you hit it, a blocking message tells you when you can use Claude again: "5-hour limit reached - resets [time]." Paid users with usage credits enabled in Usage settings see a different message instead, "5-hour limit resets [time] - continuing with usage credits.", and are not blocked. The help center notes this appears only for members with access to usage credits. The window is the same one Claude Code uses: usage of all Claude surfaces (claude.ai, Claude Code, Claude Desktop) counts toward the same usage limit, the session limit resets every five hours, and Pro and Max also have a weekly limit that resets at a fixed time each week assigned to your account. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: Settings > Usage on claude.ai (session used and time remaining) - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: Settings > Usage on claude.ai (weekly bar, with a separate bar for Opus) What to do when you see it 1. Read the time in the message: That is when your five-hour session allowance refreshes. It is your own clock, not a site-wide event. 2. Open Settings > Usage: Anthropic's usage-limit article says it shows how much of the five-hour session limit you have used, the time remaining in the session, and when the weekly limit resets. 3. Turn on usage credits to keep going: With usage credits enabled, the message becomes "5-hour limit resets [time] - continuing with usage credits." and you are not blocked. Credits do not change the reset timing. 4. Upgrade for more allowance: Max 5x and Max 20x include five and twenty times the Pro per-session allowance. When does it reset? Your own window: the help center states that your session-based usage limit will reset every five hours, and that usage credits do not affect this reset timing. The time in the message is that moment; put it on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: What is the difference between "Approaching 5-hour limit." and "5-hour limit reached"? A: The first is a warning shown as you near your plan's limit within a five-hour session; you can keep chatting. The second is the blocking message that tells you when you can use Claude again. Both come from the same help-center article. Q: Why does it say "continuing with usage credits" instead? A: Because usage credits are enabled in your Usage settings. The help center says paid users with usage credits see "5-hour limit resets [time] - continuing with usage credits." and that it appears only for members with access to usage credits. Q: Does using Claude Code count against the same limit? A: Yes. The help center says usage of all Claude product surfaces, claude.ai, Claude Code and Claude Desktop, counts toward the same usage limit, so heavy Claude Code use can trigger this message in the app. Q: Is there also a weekly limit? A: Yes on Pro and Max. It applies across all models and resets at a fixed time each week that is assigned to your account. Settings > Usage shows when it resets for Opus only and for all other models. Sources: 1. Claude Help Center: Troubleshoot Claude error messages — https://support.claude.com/en/articles/12466728-troubleshoot-claude-error-messages (verified 2026-09-18) 2. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 3. Claude Help Center: What is the Pro plan? — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 4. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 5. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 6. Claude Help Center: How do usage and length limits work? — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) ### 「5-hour limit reached - resets {time}.」 — Claude URL: https://didreset.com/zh/errors/claude-5-hour-limit-reached-resets Language: zh · Surface: 网页 · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: 5-hour limit reached - resets {time}. Variants: Approaching 5-hour limit. | 5-hour limit resets [time] - continuing with usage credits. 「5-hour limit reached - resets {time}.」是 Claude 网页版与 App 在 5 小时会话额度用完时的拦截提示，帮助中心把它列在预警「Approaching 5-hour limit.」之后。Anthropic 说明会话限额每 5 小时刷新，文案里的时间是你自己的时钟，Settings > Usage 能看到。 「5-hour limit reached - resets {time}.」是什么意思？ Anthropic 的《Troubleshoot Claude error messages》描述了 Claude App 里的三条限额文案。在 5 小时会话内接近上限时显示「Approaching 5-hour limit.」；撞上后显示拦截提示，告诉你何时可以再用：「5-hour limit reached - resets [time].」 在 Usage 设置里开通了 usage credits 的付费用户看到的是另一句「5-hour limit resets [time] - continuing with usage credits.」，不会被拦。帮助中心注明这句只对有 usage credits 权限的成员显示。 这个窗口和 Claude Code 用的是同一个：所有 Claude 界面（claude.ai、Claude Code、Claude Desktop）的用量计入同一个限额，会话限额每 5 小时刷新，Pro 与 Max 还有在分配给账号的固定时刻每周刷新的周限额。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: claude.ai 的 Settings > Usage（已用比例与剩余时间） - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: claude.ai 的 Settings > Usage（周进度条，Opus 另有一条） 看到它该怎么办 1. 读文案里的时间: 那就是你 5 小时会话额度刷新的时刻。它是你自己的时钟，不是全站事件。 2. 打开 Settings > Usage: Anthropic 的额度文章说明它显示 5 小时会话限额已用多少、会话剩余时间，以及周限额何时刷新。 3. 开通 usage credits 继续用: 开通后文案变成「5-hour limit resets [time] - continuing with usage credits.」，不再被拦。credits 不改变刷新时间。 4. 升级扩大额度: Max 5x 与 Max 20x 分别是 Pro 每会话额度的 5 倍与 20 倍。 什么时候重置？ 你自己的窗口：帮助中心写明会话额度每 5 小时刷新，且 usage credits 不影响这个刷新时间。文案里的时间就是那个时刻；可以用下方链接的窗口倒计时记下它。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: 「Approaching 5-hour limit.」和「5-hour limit reached」有什么区别？ A: 前者是在 5 小时会话内接近计划上限时的预警，还能继续聊；后者是拦截提示，告诉你何时能再用 Claude。两句都来自同一篇帮助中心文章。 Q: 为什么我看到的是「continuing with usage credits」？ A: 因为你在 Usage 设置里开通了 usage credits。帮助中心说明开通的付费用户看到的是「5-hour limit resets [time] - continuing with usage credits.」，而且只对有 usage credits 权限的成员显示。 Q: 用 Claude Code 也算进同一个限额吗？ A: 算。帮助中心写明所有 Claude 产品界面（claude.ai、Claude Code、Claude Desktop）的用量计入同一个限额，所以在 Claude Code 里用得多，App 里也会出现这句。 Q: 还有周限额吗？ A: Pro 与 Max 有。它跨所有模型计算，在分配给你账号的固定时刻每周刷新。Settings > Usage 会分别显示 Opus 与其他模型的周刷新时间。 Sources: 1. Claude Help Center: Troubleshoot Claude error messages — https://support.claude.com/en/articles/12466728-troubleshoot-claude-error-messages (verified 2026-09-18) 2. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 3. Claude Help Center: What is the Pro plan? — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 4. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 5. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 6. Claude Help Center: How do usage and length limits work? — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) ### "You've hit your limit for Claude messages. Limits will reset at {H:MM AM}. View your usage details." — Claude URL: https://didreset.com/errors/claude-you-have-hit-your-limit-for-claude-messages-limits-will-reset-at Language: en · Surface: Web · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your limit for Claude messages. Limits will reset at {H:MM AM}. View your usage details. Variants: You’ve hit your limit for Claude messages. Limits will reset at 9:30 AM. View your usage details. "You've hit your limit for Claude messages. Limits will reset at {H:MM AM}. View your usage details." is the claude.ai chat message when your allowance is used up; "View your usage details" links to claude.ai/settings/usage. The time is your own reset for the five-hour session or the weekly window, which Anthropic says reset every five hours and at a fixed weekly time. Issue #80062 shows a Max user still blocked past the time. It is not an account-wide reset. What does "You've hit your limit for Claude messages. Limits will reset at {H:MM AM}. View your usage details." mean? This is the limit dialog in the claude.ai web app. The "View your usage details" link opens claude.ai/settings/usage, where Anthropic's usage-limit article says you can see how much of the five-hour session limit you have used, the time remaining, and when the weekly limit resets for Opus only and for all other models. Issue #80062 (2026-07-22) quotes it verbatim from a Max plan account: "You’ve hit your limit for Claude messages. Limits will reset at 9:30 AM. View your usage details." The reporter was blocked on typing "hello" and saw the same 9:30 AM after waiting more than five hours; the issue was closed as not planned because it concerned claude.ai rather than the Claude Code repository. The wording does not say which window is full. Pro and Max have a session limit that resets every five hours and a weekly limit that resets at a fixed time each week assigned to your account; a reset time hours away points to the session window, a day or more away to the weekly one. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: Settings > Usage on claude.ai (session used and time remaining) - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: Settings > Usage on claude.ai (weekly bar, with a separate bar for Opus) What to do when you see it 1. Click "View your usage details": It opens claude.ai/settings/usage. Read the session bar, the time remaining in the session, and the weekly reset times before deciding what to do. 2. Wait for the time in the message: It is your own reset moment. The session allowance refreshes every five hours; the weekly one at a fixed weekly time. 3. Refresh and check again if the time has passed: Issue #80062 reports the same 9:30 AM still showing after more than five hours. Confirm on Settings > Usage whether the bar has actually cleared. 4. Contact support if you stay blocked: If Settings > Usage shows room and the dialog persists, use the help center's troubleshooting article and support channels rather than the Claude Code issue tracker, which closed #80062 as not planned. When does it reset? Your own window: the time in the message is when your window refreshes. Anthropic states that the session limit resets every five hours and the weekly limit at a fixed time each week that is assigned to your account. Settings > Usage shows both. Put the time on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: Is this the five-hour limit or the weekly limit? A: The message does not say. Open claude.ai/settings/usage via "View your usage details": it shows the five-hour session bar with time remaining and the weekly reset times. A reset a few hours away is the session window; a day or more away is the weekly window. Q: The reset time passed and I am still blocked. Why? A: Issue #80062 describes exactly that on a Max plan: the same 9:30 AM after more than five hours. Check Settings > Usage to see whether the bar cleared, then reload. If it did not, the help center's troubleshooting article and support are the route; the Claude Code repository closed the issue as not planned. Q: Does Claude Code usage count toward this? A: Yes. Both Pro and Max plans offer usage limits that are shared across Claude and Claude Code, so activity in either tool counts against the same limits. Heavy Claude Code use can produce this dialog in the web app. Q: Where does "View your usage details" go? A: To claude.ai/settings/usage. Anthropic's usage-limit article lists what it shows: how much of the five-hour session limit you have used, the time remaining in the session, and when the weekly limit resets for Opus only and for all other models. Sources: 1. anthropics/claude-code issue #80062 (2026-07-22): My claude code keeps on saying You've hit your limit for Claude messages. Limits will reset at 9:30 AM. (Max plan, claude.ai web) — https://github.com/anthropics/claude-code/issues/80062 (verified 2026-09-17) 2. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 3. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 4. Claude Help Center: What is the Pro plan? — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 5. Claude Help Center: Troubleshoot Claude error messages — https://support.claude.com/en/articles/12466728-troubleshoot-claude-error-messages (verified 2026-09-18) 6. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) ### 「You've hit your limit for Claude messages. Limits will reset at {H:MM AM}. View your usage details.」 — Claude URL: https://didreset.com/zh/errors/claude-you-have-hit-your-limit-for-claude-messages-limits-will-reset-at Language: zh · Surface: 网页 · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've hit your limit for Claude messages. Limits will reset at {H:MM AM}. View your usage details. Variants: You’ve hit your limit for Claude messages. Limits will reset at 9:30 AM. View your usage details. 「You've hit your limit for Claude messages. Limits will reset at {H:MM AM}. …」是 claude.ai 对话框在额度用完时的提示，链接指向 claude.ai/settings/usage。时间是你自己会话或周窗口的刷新时刻；前者每 5 小时刷新、后者在固定周时刻刷新。 「You've hit your limit for Claude messages. Limits will reset at {H:MM AM}. View your usage details.」是什么意思？ 这是 claude.ai 网页版的限额对话框。「View your usage details」链接打开 claude.ai/settings/usage，Anthropic 的额度文章说明那里能看到 5 小时会话限额已用多少、剩余时间，以及周限额分别对 Opus 与其他模型何时刷新。 issue #80062（2026-07-22）逐字引用了 Max 账号看到的原文：「You’ve hit your limit for Claude messages. Limits will reset at 9:30 AM. View your usage details.」报告者输入 hello 就被拦，等了 5 小时以上仍显示同一个 9:30 AM；因为问题出在 claude.ai 而非 Claude Code 仓库，issue 以 not planned 关闭。 文案不说明满的是哪个窗口。Pro 与 Max 有每 5 小时刷新的会话限额，以及在分配给账号的固定时刻每周刷新的周限额；刷新时间在几小时后是会话窗口，一天以上是周窗口。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: claude.ai 的 Settings > Usage（已用比例与剩余时间） - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: claude.ai 的 Settings > Usage（周进度条，Opus 另有一条） 看到它该怎么办 1. 点「View your usage details」: 它打开 claude.ai/settings/usage。先看会话进度条、会话剩余时间和周刷新时间，再决定怎么办。 2. 等文案里的时间: 那是你自己的刷新时刻。会话额度每 5 小时刷新；周额度在固定的周时刻刷新。 3. 时间过了就刷新页面再看: issue #80062 报告过了 5 小时以上仍显示同一个 9:30 AM。到 Settings > Usage 确认进度条是否真的清零。 4. 持续被拦就联系支持: 如果 Settings > Usage 显示有余量但对话框不消失，走帮助中心的故障排查文章和支持渠道，而不是 Claude Code 的 issue 仓库，那里把 #80062 以 not planned 关闭了。 什么时候重置？ 你自己的窗口：文案里的时间就是你窗口刷新的时刻。Anthropic 写明会话限额每 5 小时刷新、周限额在分配给你账号的固定时刻每周刷新，Settings > Usage 两者都显示。可以用下方链接的窗口倒计时记下这个时间。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: 这是 5 小时限额还是周限额？ A: 文案不说明。通过「View your usage details」打开 claude.ai/settings/usage：那里有带剩余时间的 5 小时会话进度条和周刷新时间。刷新在几小时后是会话窗口，一天以上是周窗口。 Q: 刷新时间过了还是被拦，为什么？ A: issue #80062 在 Max 计划上描述的正是这种情况：5 小时以上后仍显示同一个 9:30 AM。先到 Settings > Usage 看进度条是否清零，再重新加载。若没有，走帮助中心的故障排查文章和支持渠道；Claude Code 仓库把该 issue 以 not planned 关闭了。 Q: Claude Code 的用量也算在里面吗？ A: 算。Pro 与 Max 的额度在 Claude 与 Claude Code 之间共享，两个工具里的活动都计入同一个限额。Claude Code 用得多，网页版也会出现这个对话框。 Q: 「View your usage details」指向哪里？ A: claude.ai/settings/usage。Anthropic 的额度文章列出了它显示的内容：5 小时会话限额已用多少、会话剩余时间，以及周限额分别对 Opus 与其他模型何时刷新。 Sources: 1. anthropics/claude-code issue #80062 (2026-07-22): My claude code keeps on saying You've hit your limit for Claude messages. Limits will reset at 9:30 AM. (Max plan, claude.ai web) — https://github.com/anthropics/claude-code/issues/80062 (verified 2026-09-17) 2. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 3. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 4. Claude Help Center: What is the Pro plan? — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 5. Claude Help Center: Troubleshoot Claude error messages — https://support.claude.com/en/articles/12466728-troubleshoot-claude-error-messages (verified 2026-09-18) 6. Claude Help Center: Use Claude Code with your Pro or Max plan (/status, /login, shared limits) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) ### "Paused — you've hit your limit." — Claude Design URL: https://didreset.com/errors/claude-design-paused-you-have-hit-your-limit Language: en · Surface: App · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Paused — you've hit your limit. Variants: You've used 95% of your weekly limit · It resets Thu, Sep 10 | You're close to your weekly limit. "Paused — you've hit your limit." is the pause modal in Claude Design, the canvas in the Claude desktop app. Issue #92927 (2026-09-08) shows it on a Max 20x account at 95% of the weekly limit, with the footer "You've used 95% of your weekly limit · It resets Thu, Sep 10"; Resume worked after about 30 clicks, so the account was not actually out. The footer's reset is your own weekly time; Settings > Usage shows the real number. What does "Paused — you've hit your limit." mean? Claude Design shows this modal when it treats your plan allowance as used up. In issue #92927 the footer under the canvas read "You've used 95% of your weekly limit · It resets Thu, Sep 10" and the settings page carried a banner, "You're close to your weekly limit." The reporter, on Max 20x, was able to continue after clicking Resume about 30 times, which points to a false pause rather than an exhausted allowance. The issue was labeled invalid and closed as not planned because Claude Design is not part of the Claude Code repository; it is still the only verbatim record of the message. The limit behind the footer is the weekly one Anthropic describes for Pro and Max: it applies across all models and resets at a fixed time each week that is assigned to your account. Settings > Usage on claude.ai shows the same weekly bar and reset time. Which window is full? - Window: Session limit (5-hour) · Resets: Every five hours; the exact time is the one in the message · Where to check: Settings > Usage on claude.ai (session used and time remaining) - Window: Weekly limit · Resets: At a fixed time each week that is assigned to your account · Where to check: Settings > Usage on claude.ai (weekly bar, with a separate bar for Opus) What to do when you see it 1. Click Resume: In issue #92927 the pause lifted after repeated Resume clicks on an account at 95%, not 100%. Try it before assuming you are out. 2. Check the real percentage: Open Settings > Usage on claude.ai. Anthropic's usage-limit article says it shows the session bar with time remaining and when the weekly limit resets for Opus only and for all other models. 3. Wait for the time in the footer: If the weekly bar really is full, the footer's "It resets" date is your own weekly reset, fixed for your account each week. 4. Report it to Anthropic support: The Claude Code repository closed #92927 as invalid because Claude Design is outside its scope; use the help center for a persistent false pause. When does it reset? Your own window: the date after "It resets" in the footer is your weekly reset. Anthropic says the weekly limit resets at a fixed time each week that is assigned to your account, and the five-hour session limit resets every five hours. Put the date on a countdown with the window tracker linked below. Account-wide resets: separately, Anthropic sometimes announces a reset for everyone. Those are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled, so this site does not promise one. FAQ: Q: Am I really out of usage when I see "Paused — you've hit your limit."? A: Not necessarily. Issue #92927 documents the modal at 95% of the weekly limit on Max 20x, and Resume eventually worked. Check Settings > Usage on claude.ai for the actual percentage before waiting. Q: What does "You've used 95% of your weekly limit · It resets Thu, Sep 10" mean? A: It is the usage footer in Claude Design. The percentage is your weekly allowance used, and the date is your own weekly reset, which Anthropic says happens at a fixed time each week assigned to your account. Q: Why was the issue closed? A: It was labeled invalid and closed as not planned because Claude Design is not part of the anthropics/claude-code repository. That does not mean the behavior was denied; it means the tracker was the wrong place. Use Anthropic support for Claude Design problems. Q: Does Claude Design share limits with Claude Code and claude.ai? A: The footer in #92927 reports the same weekly limit Anthropic describes for Pro and Max, which applies across all models. Anthropic's help center says usage on claude.ai, Claude Code and Claude Desktop counts toward the same usage limit. Sources: 1. anthropics/claude-code issue #92927 (2026-09-08): Claude Design: false 'hit your limit' pause at 95% weekly on Max 20x; Resume succeeds after ~30 clicks — https://github.com/anthropics/claude-code/issues/92927 (verified 2026-09-17) 2. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 3. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 4. Claude Help Center: How do usage and length limits work? — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) ### 「Paused — you've hit your limit.」 — Claude Design URL: https://didreset.com/zh/errors/claude-design-paused-you-have-hit-your-limit Language: zh · Surface: App · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: Paused — you've hit your limit. Variants: You've used 95% of your weekly limit · It resets Thu, Sep 10 | You're close to your weekly limit. 「Paused — you've hit your limit.」是 Claude Design 的暂停弹窗。issue #92927 显示它在 Max 20x 周限额 95% 时出现，页脚写着「You've used 95% of your weekly limit …」，点 Resume 约 30 次后可继续。页脚的日期是你自己的周刷新，真实数字看 Settings。 「Paused — you've hit your limit.」是什么意思？ Claude Design 在判定你的计划额度用完时弹出这个框。issue #92927 里画布下方页脚显示「You've used 95% of your weekly limit · It resets Thu, Sep 10」，设置页还有横幅「You're close to your weekly limit.」 报告者用的是 Max 20x，点了约 30 次 Resume 后能继续，说明是误报而不是额度耗尽。该 issue 被标为 invalid 并以 not planned 关闭，因为 Claude Design 不属于 Claude Code 仓库；但它仍是这句文案唯一的逐字记录。 页脚背后的限额就是 Anthropic 为 Pro 与 Max 描述的周限额：跨所有模型计算，在分配给你账号的固定时刻每周刷新。claude.ai 的 Settings > Usage 显示同一条周进度条和刷新时间。 满的是哪个窗口？ - 窗口: 会话窗口（5 小时） · 何时刷新: 每 5 小时刷新；具体时刻以文案里的时间为准 · 在哪里看: claude.ai 的 Settings > Usage（已用比例与剩余时间） - 窗口: 周窗口 · 何时刷新: 每周在分配给你账号的固定时刻刷新 · 在哪里看: claude.ai 的 Settings > Usage（周进度条，Opus 另有一条） 看到它该怎么办 1. 点 Resume: issue #92927 里账号在 95%（不是 100%）时反复点 Resume 后恢复了。先试这一步，不要先假定额度用完。 2. 看真实百分比: 打开 claude.ai 的 Settings > Usage。Anthropic 的额度文章说明它显示带剩余时间的会话进度条，以及周限额分别对 Opus 与其他模型何时刷新。 3. 等页脚里的时间: 如果周进度条真的满了，页脚「It resets」后面的日期就是你自己的周刷新，每周对你的账号固定。 4. 向 Anthropic 支持反馈: Claude Code 仓库以 invalid 关闭了 #92927，因为 Claude Design 不在其范围内；持续误报请走帮助中心。 什么时候重置？ 你自己的窗口：页脚「It resets」后面的日期是你的周刷新。Anthropic 写明周限额在分配给你账号的固定时刻每周刷新，5 小时会话限额每 5 小时刷新。可以用下方链接的窗口倒计时记下这个日期。 全员重置：另一回事是 Anthropic 偶尔宣布面向所有人的重置。本站板面记录的就是这些公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的，本站不承诺下一次。 FAQ: Q: 看到「Paused — you've hit your limit.」就是额度真的用完了吗？ A: 不一定。issue #92927 记录了它在 Max 20x 周限额 95% 时出现，而且 Resume 最终有效。等之前先到 claude.ai 的 Settings > Usage 看真实百分比。 Q: 「You've used 95% of your weekly limit · It resets Thu, Sep 10」是什么意思？ A: 那是 Claude Design 的用量页脚。百分比是你已用的周额度，日期是你自己的周刷新，Anthropic 说明它在分配给你账号的固定时刻每周发生。 Q: 这条 issue 为什么被关了？ A: 它被标为 invalid 并以 not planned 关闭，因为 Claude Design 不属于 anthropics/claude-code 仓库。这不代表否认了该现象，只是提交的地方不对。Claude Design 的问题请走 Anthropic 支持。 Q: Claude Design 和 Claude Code、claude.ai 共用限额吗？ A: #92927 的页脚报告的就是 Anthropic 为 Pro 与 Max 描述的那条跨所有模型的周限额。Anthropic 帮助中心写明 claude.ai、Claude Code、Claude Desktop 的用量计入同一个限额。 Sources: 1. anthropics/claude-code issue #92927 (2026-09-08): Claude Design: false 'hit your limit' pause at 95% weekly on Max 20x; Resume succeeds after ~30 clicks — https://github.com/anthropics/claude-code/issues/92927 (verified 2026-09-17) 2. Claude Help Center: What is the Max plan? (session resets every five hours; weekly limit at a fixed time) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 3. Claude Help Center: Usage limit best practices (Settings > Usage) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 4. Claude Help Center: How do usage and length limits work? — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) ### "You've reached your Grok Bot usage limit. It resets in {N} days." — Grok Bot URL: https://didreset.com/errors/grok-bot-you-have-reached-your-grok-bot-usage-limit-it-resets-in-days Language: en · Surface: IDE · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've reached your Grok Bot usage limit. It resets in {N} days. Variants: You’ve reached your Grok Bot usage limit. It resets in 3 days. "You've reached your Grok Bot usage limit. It resets in {N} days." is the banner in Cursor when your weekly Grok Bot allowance is at 100%. xAI's docs say "Usage resets weekly." and that subscriptions include weekly usage with optional on-demand usage; the size of the allowance is not published. A Cursor staff reply on 2026-09-02 confirms "Your weekly usage resets on its own (the app shows the countdown)." The countdown is yours, not an account-wide reset. What does "You've reached your Grok Bot usage limit. It resets in {N} days." mean? Grok Bot draws from a weekly allowance. xAI's Grok Bot overview says "Usage resets weekly.", and its FAQ says "Grok Bot subscriptions include weekly usage; eligible accounts can add on-demand usage billed from model and token cost." When the sidebar shows weekly usage at 100%, Cursor shows this banner with a countdown in days. The verbatim record is a Cursor forum thread from 2026-09-01: the banner read "You’ve reached your Grok Bot usage limit. It resets in 3 days." and appeared twice within 21 seconds. A Cursor staff member replied on 2026-09-02 that the two banners were "just two blocked attempts after the pool was already full, not a double charge" and that "Your weekly usage resets on its own (the app shows the countdown)." xAI has not published how much weekly usage each plan includes: not published. Since 2026-08-26 Grok Bot is included with SuperGrok, SuperGrok Plus, SuperGrok Heavy and paid Cursor plans, with "its own usage, separate from your Grok and Cursor plans"; with both a Cursor and a SuperGrok subscription, "Grok Bot uses whichever has more usage." What to do when you see it 1. Read the countdown: The banner's "It resets in N days" is your own weekly reset. Cursor staff: "Your weekly usage resets on its own (the app shows the countdown)." 2. Check usage and billing: xAI's troubleshooting page says that when usage is exhausted or an on-demand spending limit is reached, review usage and billing from the account access page or Usage & Billing. 3. Add on-demand usage if you cannot wait: The Grok Bot FAQ says eligible accounts can add on-demand usage billed from model and token cost. For Grok chat plans, the Grok FAQ lists Extra Usage Credits, Auto Top Up or a higher plan. 4. Stop bots from waking each other: In the forum thread, bot-to-bot review messages each ran a turn that counted toward weekly usage. Cursor staff suggested running the work in one Command Agent with subagents and deleting unused specialist bots. 5. Link a SuperGrok subscription: If you have both a Cursor and a SuperGrok subscription, Grok Bot uses whichever has more usage, per xAI's FAQ. When does it reset? Your own window: Grok Bot usage resets weekly, on the countdown the app shows; xAI does not publish the allowance size or a fixed reset hour for Grok Bot. For Grok chat, the Grok FAQ says "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." Neither is something this site can compute for you. Account-wide resets: xAI has also reset weekly usage for everyone, for example on 2026-08-26 when @bot announced "We're also resetting weekly usage limits for all users." Those announcements are what this board records, each linked to its original post, and the row above shows the last one. They are announced, not scheduled. FAQ: Q: How much weekly Grok Bot usage do I get? A: Not published. xAI's docs say Grok Bot subscriptions include weekly usage and that usage resets weekly, but no plan-by-plan amount is stated. This site does not estimate one; the countdown in the banner is the only official signal about your own reset. Q: The banner appeared twice. Was I charged twice? A: No, according to Cursor staff in the forum thread: the second banner was "just two blocked attempts after the pool was already full, not a double charge." Q: Can I keep working before the reset? A: xAI's Grok Bot FAQ says eligible accounts can add on-demand usage billed from model and token cost, and the troubleshooting page points to Usage & Billing when usage is exhausted. For Grok chat plans, the Grok FAQ lists Extra Usage Credits, Auto Top Up or upgrading. Q: Does my SuperGrok plan add to my Cursor Grok Bot usage? A: Not additively. xAI's FAQ says that with both a Cursor and a SuperGrok subscription, "Grok Bot uses whichever has more usage." Grok Bot's allowance is also separate from your Grok and Cursor plan usage, per the 2026-08-26 announcement. Sources: 1. Cursor forum thread 170271 (2026-09-01): Grok Bot weekly usage hits 100%, with a Cursor staff reply — https://forum.cursor.com/t/grok-bot-weekly-usage-hits-100-after-bot-to-bot-reviews-the-user-asked-to-stop/170271 (verified 2026-09-18) 2. xAI docs: Grok Bot overview ("Usage resets weekly.") — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 3. xAI docs: Grok Bot FAQ (weekly usage, on-demand usage, whichever has more usage) — https://docs.x.ai/grok-bot/faq (verified 2026-09-17) 4. xAI docs: Grok FAQ (weekly usage pool, Settings → Usage, Extra Usage Credits) — https://docs.x.ai/grok/faq (verified 2026-09-18) 5. xAI news, 2026-08-26: Grok Bot is now included with more plans — https://x.ai/news/grok-bot-more-plans (verified 2026-09-17) 6. xAI docs: Grok Bot troubleshooting (usage exhausted → Usage & Billing) — https://docs.x.ai/grok-bot/troubleshooting (verified 2026-09-17) 7. Grok Bot (@bot) on X, 2026-08-26: resetting weekly usage limits for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) ### 「You've reached your Grok Bot usage limit. It resets in {N} days.」 — Grok Bot URL: https://didreset.com/zh/errors/grok-bot-you-have-reached-your-grok-bot-usage-limit-it-resets-in-days Language: zh · Surface: IDE · Published: 2026-09-18 · Updated: 2026-09-18 Exact message: You've reached your Grok Bot usage limit. It resets in {N} days. Variants: You’ve reached your Grok Bot usage limit. It resets in 3 days. 「You've reached your Grok Bot usage limit. It resets in {N} days.」是 Cursor 在 Grok Bot 周额度到 100% 时的横幅。xAI 文档写明「Usage resets weekly.」，额度大小未公布。Cursor 员工 2026-09-02 回复：周用量会自己刷新，App 显示倒计时。 「You've reached your Grok Bot usage limit. It resets in {N} days.」是什么意思？ Grok Bot 从周额度里扣用量。xAI 的 Grok Bot 概览写「Usage resets weekly.」，FAQ 写「Grok Bot subscriptions include weekly usage; eligible accounts can add on-demand usage billed from model and token cost.」侧栏显示周用量 100% 时，Cursor 就显示这条带天数倒计时的横幅。 逐字记录来自 2026-09-01 的 Cursor 论坛帖：横幅原文是「You’ve reached your Grok Bot usage limit. It resets in 3 days.」，21 秒内出现两次。Cursor 员工 2026-09-02 回复，两次横幅是「just two blocked attempts after the pool was already full, not a double charge」，并且「Your weekly usage resets on its own (the app shows the countdown).」 xAI 没有公布各计划的周用量是多少：未公布。2026-08-26 起 Grok Bot 随 SuperGrok、SuperGrok Plus、SuperGrok Heavy 与 Cursor 付费计划提供，「its own usage, separate from your Grok and Cursor plans」；同时有 Cursor 与 SuperGrok 订阅时，「Grok Bot uses whichever has more usage.」 看到它该怎么办 1. 读倒计时: 横幅里的「It resets in N days」是你自己的周刷新。Cursor 员工：「Your weekly usage resets on its own (the app shows the countdown).」 2. 看用量与账单: xAI 的故障排查页写明，用量耗尽或按需消费达到上限时，到账号访问页或 Usage & Billing 查看用量与账单。 3. 等不及就加购按需用量: Grok Bot FAQ 写明符合条件的账号可以按模型与 token 成本加购按需用量。Grok 聊天计划则按 Grok FAQ：Extra Usage Credits、Auto Top Up 或升级更高计划。 4. 别让 Bot 互相唤醒: 论坛帖里 Bot 之间的评审消息每一条都跑一轮、都计入周用量。Cursor 员工建议在一个 Command Agent 里用子代理完成工作，并删除不用的专项 Bot。 5. 关联 SuperGrok 订阅: 按 xAI FAQ，同时有 Cursor 与 SuperGrok 订阅时，Grok Bot 用用量更多的那一个。 什么时候重置？ 你自己的窗口：Grok Bot 用量每周刷新，以 App 里显示的倒计时为准；xAI 没有公布 Grok Bot 的额度大小或固定的刷新时刻。Grok 聊天则按 Grok FAQ：「The usage pool limit resets every week on a schedule shown in the Usage tab in Settings.」两者本站都无法替你计算。 全员重置：xAI 也给所有人重置过周用量，例如 2026-08-26 @bot 宣布「We're also resetting weekly usage limits for all users.」本站板面记录的就是这类公告，每条都链接原帖，上方板面行显示最近一次。它们是宣布的、不是排期的。 FAQ: Q: Grok Bot 每周有多少用量？ A: 未公布。xAI 文档只说 Grok Bot 订阅自带周用量、用量每周刷新，没有按计划给出数字。本站不估算；横幅里的倒计时是关于你自己刷新的唯一官方信号。 Q: 横幅出现了两次，是扣了两次费吗？ A: 不是。论坛帖里 Cursor 员工说明，第二次横幅「just two blocked attempts after the pool was already full, not a double charge」。 Q: 刷新之前还能继续用吗？ A: xAI 的 Grok Bot FAQ 写明符合条件的账号可以按模型与 token 成本加购按需用量，故障排查页在用量耗尽时指向 Usage & Billing。Grok 聊天计划则按 Grok FAQ：Extra Usage Credits、Auto Top Up 或升级。 Q: 我的 SuperGrok 计划会叠加到 Cursor 的 Grok Bot 用量上吗？ A: 不叠加。xAI FAQ 写明同时有 Cursor 与 SuperGrok 订阅时，「Grok Bot uses whichever has more usage.」而且按 2026-08-26 的公告，Grok Bot 的额度与 Grok、Cursor 计划的用量是分开的。 Sources: 1. Cursor forum thread 170271 (2026-09-01): Grok Bot weekly usage hits 100%, with a Cursor staff reply — https://forum.cursor.com/t/grok-bot-weekly-usage-hits-100-after-bot-to-bot-reviews-the-user-asked-to-stop/170271 (verified 2026-09-18) 2. xAI docs: Grok Bot overview ("Usage resets weekly.") — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 3. xAI docs: Grok Bot FAQ (weekly usage, on-demand usage, whichever has more usage) — https://docs.x.ai/grok-bot/faq (verified 2026-09-17) 4. xAI docs: Grok FAQ (weekly usage pool, Settings → Usage, Extra Usage Credits) — https://docs.x.ai/grok/faq (verified 2026-09-18) 5. xAI news, 2026-08-26: Grok Bot is now included with more plans — https://x.ai/news/grok-bot-more-plans (verified 2026-09-17) 6. xAI docs: Grok Bot troubleshooting (usage exhausted → Usage & Billing) — https://docs.x.ai/grok-bot/troubleshooting (verified 2026-09-17) 7. Grok Bot (@bot) on X, 2026-08-26: resetting weekly usage limits for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) ## Guides: rules, how-to and glossary ### Does the Codex 5-hour window reset on a fixed clock? URL: https://didreset.com/guides/codex-5-hour-window Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 No. OpenAI's help center says 'A new window starts when you send your first message in Work or Codex after the previous window ends', so the clock is per account, not a fixed schedule. Local CLI messages and cloud chats share one allowance, and clearing the window does not refill the weekly limit. Since 2026-08-25 the five-hour limit applies to Plus but not Pro. Run /status in the CLI to see your own reset time. Definition Codex meters subscription usage in two windows. The help center says usage limits 'may apply over a five-hour window and a weekly window' depending on your plan, and 'Where both apply, you need allowance remaining in both to continue'. Hitting the five-hour cap does not touch the weekly clock, and you can hit it 'before five hours have passed, even if you still have weekly usage remaining'. The pool is shared across surfaces. The pricing page states 'Local messages and cloud chats share your plan's usage allowance', and the help center adds that Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents 'use a shared allowance and credit pool when those features are available on your plan'. OpenAI publishes per-plan five-hour estimates, for example 5–45 GPT-6 Astra local messages on Plus, but calls them 'not fixed message limits'. Which plans have the window has changed. On 2026-07-12 Tibo Sottiaux announced Codex was 'Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans'. In a post timestamped 2026-08-25 01:16 UTC he wrote 'Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex'; read in Pacific time that puts the return on 2026-08-25, and the same post keeps the limit off for Pro. This site dates the return by the post, not by a separate announcement. What the official docs say - "A new window starts when you send your first message in Work or Codex after the previous window ends." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Where both apply, you need allowance remaining in both to continue" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Local messages and cloud chats share your plan's usage allowance. Weekly limits may also apply." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans" — Tibo Sottiaux on X, 2026-07-12 (5-hour limit removed for Plus, Business and Pro) (https://x.com/thsottiaux/status/2076365965915467978) - "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex." — Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") (https://x.com/thsottiaux/status/2092058556707344708) How it is counted | Question | Codex answer | Source | | --- | --- | --- | | When does a new 5-hour window start? | With your first message in Work or Codex after the previous window ends | [1] | | Is it a fixed clock? | No. It is per account and starts from that first message | [1] | | Does clearing it refill the weekly limit? | No. Where both apply you need allowance remaining in both | [1] | | Do CLI, IDE and cloud share it? | Yes. Local messages and cloud chats share one plan allowance; ChatGPT Work shares it too | [2] [3] | | Which plans have it? | Plus since 2026-08-25; Pro $100 and $200 not enabled; Business not named in the return post | [4] [5] | | Published number? | Estimates only, e.g. 5–45 GPT-6 Astra local messages per five hours on Plus; "not fixed message limits" | [2] | FAQ: Q: Can I hit the Codex 5-hour limit before five hours have passed? A: Yes. OpenAI's help center says 'You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining.' The window is an allowance that starts with your first message, not a timer that guarantees five hours of work. Q: Does the 5-hour window reset the weekly limit? A: No. The help center says that where both windows apply 'you need allowance remaining in both to continue'. A fresh five-hour window does not add anything to the weekly allowance; only the weekly clock, a banked reset, a purchased reset or an official account-wide reset does. Q: Does the Codex 5-hour limit apply to Pro? A: Not as of the last check. Tibo Sottiaux's post of 2026-08-25 01:16 UTC brought the 5h limit back for Plus the next day and said Codex was 'keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions' for the upcoming months. Business plans were not named in that post. Q: Where do I see when my own 5-hour window resets? A: Type /status in an active Codex CLI session; the pricing page says that is where you see your remaining limits. The desktop app and web show the same windows and reset times under Settings → Usage. Q: Is this the same as a reset on the DidReset board? A: No. Your 5-hour window refreshes on its own per-account clock. The board records account-wide resets announced by OpenAI, which refresh everyone at once and are the only resets this site tracks. Sources: 1. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 3. OpenAI Help Center: Using Codex with your ChatGPT plan — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-07-12 (5-hour limit removed for Plus, Business and Pro) — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 5. Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### Codex 的 5 小时窗口是按固定时钟重置的吗？ URL: https://didreset.com/zh/guides/codex-5-hour-window Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 不是。OpenAI 帮助中心写明：上一个窗口结束后，你在 Work 或 Codex 发出第一条消息时才开始新的 5 小时窗口，时钟按账号各自计算。本地 CLI 消息与云端会话共用同一份额度；5 小时窗口刷新也不会补回周限额。自 2026-08-25 起 5 小时限制只对 Plus 生效，Pro 不受限。在 CLI 里输入 /status 可看到自己的重置时间。 定义 Codex 订阅用量分两个窗口计量。帮助中心说，视计划而定，限额可能同时按 5 小时窗口与周窗口计算；两者都适用时，两边都要有剩余额度才能继续。5 小时窗口满了不影响周时钟，而且即使周额度还有剩余，也可能不到 5 小时就先碰到 5 小时上限。 额度池跨入口共用。定价页写明本地消息与云端会话共用计划额度；帮助中心补充说 Codex、ChatGPT Work、ChatGPT for Excel 与 Workspace Agents 在计划包含这些功能时使用同一个额度与积分池。OpenAI 公布了各计划的 5 小时估算值（例如 Plus 每 5 小时约 5–45 条 GPT-6 Astra 本地消息），但明确说这不是固定的消息条数上限。 哪些计划有这个窗口是变过的。2026-07-12，Codex 负责人 Tibo Sottiaux 宣布暂时移除 Plus、Business、Pro 三档的 5 小时限制。他在一条发布时间为 2026-08-25 01:16 UTC 的帖子里写道「Tomorrow」将为 Plus 账号恢复 5 小时限制，按太平洋时间理解即 2026-08-25 恢复；同一条帖子说 Pro 继续不启用。本站以该帖时间为准标注恢复日期，没有另外的官方公告。 官方文档怎么说 - "A new window starts when you send your first message in Work or Codex after the previous window ends." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Where both apply, you need allowance remaining in both to continue" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Local messages and cloud chats share your plan's usage allowance. Weekly limits may also apply." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans" — Tibo Sottiaux on X, 2026-07-12 (5-hour limit removed for Plus, Business and Pro) (https://x.com/thsottiaux/status/2076365965915467978) - "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex." — Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") (https://x.com/thsottiaux/status/2092058556707344708) 怎么算 | 问题 | Codex 的规则 | 来源 | | --- | --- | --- | | 新的 5 小时窗口何时开始？ | 上一个窗口结束后，你在 Work 或 Codex 发出第一条消息时 | [1] | | 是固定时钟吗？ | 不是。按账号各自计算，从那条消息开始算 | [1] | | 5 小时窗口刷新会补回周限额吗？ | 不会。两者都适用时，两边都要有剩余额度 | [1] | | CLI、IDE、云端共用吗？ | 共用。本地消息与云端会话共用计划额度，ChatGPT Work 也在同一池 | [2] [3] | | 哪些计划有它？ | Plus 自 2026-08-25 起有；Pro $100 与 $200 未启用；Business 在恢复帖里没被点名 | [4] [5] | | 公布过具体数字吗？ | 只有估算值，例如 Plus 每 5 小时约 5–45 条 GPT-6 Astra 本地消息；官方称不是固定条数上限 | [2] | FAQ: Q: 不到 5 小时也可能碰到 Codex 的 5 小时上限吗？ A: 可能。OpenAI 帮助中心写明，即使周额度还有剩余，你也可能在 5 小时过去之前就碰到 5 小时上限。这个窗口是从第一条消息开始计的一份额度，不是保证你能连续工作 5 小时的计时器。 Q: 5 小时窗口刷新会顺便重置周限额吗？ A: 不会。帮助中心说两个窗口都适用时，两边都要有剩余额度才能继续。新的 5 小时窗口不会给周额度加任何东西；只有周时钟走到、用掉重置卡、购买即时重置，或者官方全员重置才会补回周额度。 Q: Codex 的 5 小时限制对 Pro 生效吗？ A: 截至最近一次核实不生效。Tibo Sottiaux 在 2026-08-25 01:16 UTC 的帖子里说次日为 Plus 恢复 5 小时限制，同时在接下来几个月对 Pro $100 与 Pro $200 订阅继续不启用。那条帖子没有提到 Business。 Q: 在哪里能看到自己的 5 小时窗口几点刷新？ A: 在正在运行的 Codex CLI 会话里输入 /status；定价页说那里能看到你的剩余限额。桌面应用与网页版在「设置 → 用量」（Settings → Usage）里显示同样的窗口与重置时间。 Q: 这和 DidReset 出发板上的重置是一回事吗？ A: 不是。你的 5 小时窗口按自己账号的时钟刷新。出发板记录的是 OpenAI 公告的全员重置，一次刷新所有人的额度，本站只追踪这一种。 Sources: 1. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 3. OpenAI Help Center: Using Codex with your ChatGPT plan — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-07-12 (5-hour limit removed for Plus, Business and Pro) — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 5. Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### How the Codex weekly limit is counted URL: https://didreset.com/guides/codex-weekly-limit Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 OpenAI has not published this. The Codex pricing page says only 'Weekly limits may also apply' and points you to the usage dashboard for reset times. What is documented: the weekly limit is a second cap on top of the five-hour window, tracked per account; a full banked reset 'changes your weekly reset date'; after a purchased reset the next automatic weekly reset is seven days after your first request. Type /status in the CLI to see your own date. Definition The weekly limit is the second of Codex's two windows. The help center says it 'controls how much included work you can use over the weekly usage period', and that where both windows apply 'you need allowance remaining in both to continue'. OpenAI publishes five-hour estimates per plan but no weekly figures; the pricing page's only statement is 'Weekly limits may also apply.' Your reset time is per account and is shown in the usage dashboard, in Settings → Usage, or by entering /status in an active Codex CLI session. Two things move it: 'Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date', and after a purchased reset 'your next automatic weekly reset is seven days after that request'. Enterprise and Edu workspaces on flexible pricing are different: the pricing page says they have 'no fixed rate limits—usage scales with credits'. Everything on this page is about Plus, Pro and Business subscriptions. What the official docs say - "Weekly limits may also apply." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "These estimates are not fixed message limits; check your usage dashboard for current limits and reset times." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "controls how much included work you can use over the weekly usage period" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "your next automatic weekly reset is seven days after that request" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "In an active Codex CLI session, enter /status." — OpenAI Help Center: Using Codex with your ChatGPT plan (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) How it is counted | Question | Codex answer | Source | | --- | --- | --- | | Is there a published weekly number? | Not published. The pricing page says only "Weekly limits may also apply" | [1] | | Where is my weekly reset time? | Usage dashboard, Settings → Usage, or /status in an active CLI session | [1] [4] | | What does the weekly limit control? | How much included work you can use over the weekly usage period | [2] | | Does a fresh 5-hour window refill it? | No. Where both apply you need allowance remaining in both | [2] | | After a full banked reset? | Both windows refresh and your weekly reset date changes | [3] | | After a purchased reset? | A new weekly period starts with your first request; the next automatic reset is seven days after it | [2] | | Enterprise / Edu on flexible pricing? | No fixed rate limits; usage scales with credits | [1] | FAQ: Q: What is the Codex weekly limit on Plus or Pro? A: OpenAI has not published a number for any plan. The pricing page gives five-hour estimates only and says 'Weekly limits may also apply'; it points you to the usage dashboard for 'current limits and reset times'. This site does not estimate the weekly cap. Q: When does my Codex weekly limit reset? A: On your own account clock, not on a shared weekday. The exact time is shown in the usage dashboard, in Settings → Usage, or by typing /status in an active CLI session. Paste that time into the window tracker on this site for a countdown. Q: Does a banked reset move my weekly reset date? A: Yes. The help center says a full banked reset 'refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date', and tells you to check Settings → Usage for the updated time. Q: Is the weekly reset the same as an official reset on the board? A: No. Your weekly reset happens on your own schedule and refreshes only your account. The resets on the DidReset board are announced by OpenAI and refresh everyone at once; they do not change how your weekly window is counted afterwards. Sources: 1. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 2. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. OpenAI Help Center: How banked Codex resets work — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 4. OpenAI Help Center: Using Codex with your ChatGPT plan — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) ### Codex 的周限额怎么算 URL: https://didreset.com/zh/guides/codex-weekly-limit Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 OpenAI 未公布。Codex 定价页只说周限额可能同样适用，并让你到用量面板查看重置时间。已公开的规则是：周限额是叠在 5 小时窗口之上的第二道上限，按账号各自计时；用掉一张完整重置卡会改变你的周重置日期；购买即时重置后，下一次自动周重置落在你第一次请求之后的第 7 天。CLI 里输入 /status 可看到自己的日期。 定义 周限额是 Codex 两个窗口中的第二个。帮助中心说它控制的是在一个周用量周期内可以使用多少包含在计划里的工作量；两个窗口都适用时，两边都要有剩余额度才能继续。OpenAI 只公布各计划的 5 小时估算值，没有周数字；定价页的唯一表述是周限额可能同样适用。 你的周重置时间按账号各自计算，显示在用量面板、Settings → Usage，或者在正在运行的 Codex CLI 会话里输入 /status。两件事会改变它：用掉一张完整重置卡会同时刷新 5 小时与周窗口并改变周重置日期；购买即时重置后，下一次自动周重置落在你第一次请求之后的第 7 天。 走灵活计费的 Enterprise 与 Edu 工作区不一样：定价页说它们没有固定的速率限制，用量按积分伸缩。本页其余内容都只针对 Plus、Pro、Business 订阅。 官方文档怎么说 - "Weekly limits may also apply." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "These estimates are not fixed message limits; check your usage dashboard for current limits and reset times." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "controls how much included work you can use over the weekly usage period" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "your next automatic weekly reset is seven days after that request" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "In an active Codex CLI session, enter /status." — OpenAI Help Center: Using Codex with your ChatGPT plan (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) 怎么算 | 问题 | Codex 的规则 | 来源 | | --- | --- | --- | | 公布过周限额数字吗？ | 未公布。定价页只说周限额可能同样适用 | [1] | | 我的周重置时间在哪里看？ | 用量面板、Settings → Usage，或在运行中的 CLI 会话里输入 /status | [1] [4] | | 周限额控制什么？ | 一个周用量周期内可以使用多少包含在计划里的工作量 | [2] | | 新的 5 小时窗口会补回它吗？ | 不会。两者都适用时，两边都要有剩余额度 | [2] | | 用掉完整重置卡之后？ | 两个窗口都刷新，周重置日期随之改变 | [3] | | 购买即时重置之后？ | 新的周周期从你的第一次请求开始，下一次自动重置在其 7 天之后 | [2] | | 灵活计费的 Enterprise / Edu？ | 没有固定速率限制，用量按积分伸缩 | [1] | FAQ: Q: Codex Plus 或 Pro 的周限额是多少？ A: OpenAI 没有为任何计划公布数字。定价页只给出 5 小时估算值，并说周限额可能同样适用，让你到用量面板查看当前限额与重置时间。本站不估算周上限。 Q: 我的 Codex 周限额什么时候重置？ A: 按你自己账号的时钟，不是所有人共用的某个星期几。确切时间显示在用量面板、Settings → Usage，或在运行中的 CLI 会话里输入 /status。把那个时间贴进本站的窗口倒计时工具就能倒数。 Q: 重置卡会改变我的周重置日期吗？ A: 会。帮助中心写明，用掉一张完整重置卡会刷新 5 小时与周两个 Codex 窗口并改变周重置日期，并让你到 Settings → Usage 查看更新后的时间。 Q: 周重置和出发板上的官方重置是一回事吗？ A: 不是。你的周重置按自己的时间表进行，只刷新你的账号。DidReset 出发板上的重置由 OpenAI 公告，一次刷新所有人；它不会改变此后你周窗口的计算方式。 Sources: 1. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 2. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. OpenAI Help Center: How banked Codex resets work — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 4. OpenAI Help Center: Using Codex with your ChatGPT plan — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) ### Does the Codex 5-hour limit apply to Pro plans? URL: https://didreset.com/guides/codex-5-hour-limit-pro-vs-plus Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 No. On 2026-07-12 Tibo Sottiaux announced Codex was 'Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans'. In a post timestamped 2026-08-25 01:16 UTC he said the 5h limit would return for Plus 'Tomorrow', which in Pacific time is 2026-08-25, while 'keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions' for the upcoming months. OpenAI's help center therefore says limits 'may apply' over a five-hour window depending on your plan. Definition The removal came first. On 2026-07-12 Tibo Sottiaux, who runs Codex, posted that the team was 'Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans'. Two days later he confirmed 'we continue to not have the 5h rate limit as well'. The return was partial. His post of 2026-08-25 01:16 UTC said 'Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex', explaining that 'the 5h limit allows us to smoothen the load on our compute, allowing to keep the plan generous in terms of weekly usage'. The same post said Codex was 'keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions' for the upcoming months. This site dates the Plus return to 2026-08-25 because the post was written on the evening of 2026-08-24 Pacific time; no separate announcement names the day. Business plans were named in the removal but not in the return post, so their status is unconfirmed here. The help center's wording covers all cases: 'Depending on your plan, usage limits may apply over a five-hour window and a weekly window.' What the official docs say - "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans" — Tibo Sottiaux on X, 2026-07-12 (5-hour limit removed for Plus, Business and Pro) (https://x.com/thsottiaux/status/2076365965915467978) - "And we continue to not have the 5h rate limit as well" — Tibo Sottiaux on X, 2026-07-14 (5h limit still off) (https://x.com/thsottiaux/status/2077114635308986427) - "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex." — Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") (https://x.com/thsottiaux/status/2092058556707344708) - "We are for the upcoming months keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions." — Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") (https://x.com/thsottiaux/status/2092058556707344708) - "the 5h limit allows us to smoothen the load on our compute, allowing to keep the plan generous in terms of weekly usage" — Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") (https://x.com/thsottiaux/status/2092058556707344708) - "Depending on your plan, usage limits may apply over a five-hour window and a weekly window." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) How it is counted | Plan | 5-hour limit | Since | Source | | --- | --- | --- | --- | | Plus | Applies | 2026-08-25, dated by the post of 2026-08-25 01:16 UTC that said "Tomorrow" | [3] | | Pro $100 / Pro $200 | Not enabled "for the upcoming months" | Removed 2026-07-12; exemption confirmed in the 2026-08-25 post | [1] [3] | | Business | Removed 2026-07-12; not named in the return post | Unconfirmed | [1] [3] | | Any plan | Weekly limit may still apply on top | Ongoing | [4] | FAQ: Q: Why did OpenAI bring the 5-hour limit back for Plus? A: Tibo Sottiaux's post of 2026-08-25 01:16 UTC says 'the 5h limit allows us to smoothen the load on our compute, allowing to keep the plan generous in terms of weekly usage'. The five-hour window spreads demand; the weekly allowance is what it protects. Q: Is the Pro exemption permanent? A: No end date has been published. The post says Codex is 'for the upcoming months keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions'. Treat it as current policy that can change; this page is updated when a new post changes it. Q: What about Business plans? A: Business was included in the 2026-07-12 removal but was not named in the 2026-08-25 return post. This site does not infer either way; the row above is marked unconfirmed until OpenAI says so. Q: Do Pro plans still have a weekly limit? A: The help center says 'Depending on your plan, usage limits may apply over a five-hour window and a weekly window.' Only the five-hour window was switched off for Pro; nothing in the posts says the weekly limit was removed, and OpenAI publishes no weekly number. Sources: 1. Tibo Sottiaux on X, 2026-07-12 (5-hour limit removed for Plus, Business and Pro) — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 2. Tibo Sottiaux on X, 2026-07-14 (5h limit still off) — https://x.com/thsottiaux/status/2077114635308986427 (verified 2026-09-17) 3. Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) 4. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) ### Codex 的 5 小时限制适用于 Pro 档吗？ URL: https://didreset.com/zh/guides/codex-5-hour-limit-pro-vs-plus Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 不适用。2026-07-12，Tibo Sottiaux 宣布 Codex 暂时移除 Plus、Business、Pro 的 5 小时限制。他在 2026-08-25 01:16 UTC 的帖子里说「Tomorrow」为 Plus 恢复，即太平洋时间 2026-08-25，而 Pro 接下来几个月继续不启用。帮助中心因此只写限额可能按 5 小时窗口计算。 定义 先是移除。2026-07-12，Codex 负责人 Tibo Sottiaux 发帖说团队暂时移除 Plus、Business、Pro 三档的 5 小时用量限制。两天后他确认 5 小时限制仍未启用。 之后是部分恢复。他在 2026-08-25 01:16 UTC 发布的帖子说，次日将为 ChatGPT Work 与 Codex 的 Plus 账号恢复 5 小时限制，理由是 5 小时限制能平滑算力负载，从而让计划在周用量上保持宽松。同一条帖子说接下来几个月对 Pro $100 与 Pro $200 订阅继续不启用。本站把 Plus 的恢复日期记为 2026-08-25，依据是该帖写于太平洋时间 2026-08-24 傍晚，「Tomorrow」即 08-25；没有另外的公告点名日期。 Business 在移除时被点名，在恢复帖里没有出现，所以本页把它标为未确认。帮助中心的措辞覆盖所有情况：视计划而定，限额可能按 5 小时窗口和周窗口计算。 官方文档怎么说 - "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans" — Tibo Sottiaux on X, 2026-07-12 (5-hour limit removed for Plus, Business and Pro) (https://x.com/thsottiaux/status/2076365965915467978) - "And we continue to not have the 5h rate limit as well" — Tibo Sottiaux on X, 2026-07-14 (5h limit still off) (https://x.com/thsottiaux/status/2077114635308986427) - "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex." — Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") (https://x.com/thsottiaux/status/2092058556707344708) - "We are for the upcoming months keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions." — Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") (https://x.com/thsottiaux/status/2092058556707344708) - "the 5h limit allows us to smoothen the load on our compute, allowing to keep the plan generous in terms of weekly usage" — Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") (https://x.com/thsottiaux/status/2092058556707344708) - "Depending on your plan, usage limits may apply over a five-hour window and a weekly window." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) 怎么算 | 计划 | 5 小时限制 | 起自 | 来源 | | --- | --- | --- | --- | | Plus | 生效 | 2026-08-25，依据 2026-08-25 01:16 UTC 那条写着「Tomorrow」的帖子 | [3] | | Pro $100 / Pro $200 | 接下来几个月不启用 | 2026-07-12 移除；2026-08-25 的帖子确认继续豁免 | [1] [3] | | Business | 2026-07-12 移除；恢复帖未点名 | 未确认 | [1] [3] | | 所有计划 | 周限额可能仍然叠加在上面 | 持续 | [4] | FAQ: Q: OpenAI 为什么给 Plus 恢复 5 小时限制？ A: Tibo Sottiaux 在 2026-08-25 01:16 UTC 的帖子里说，5 小时限制能平滑算力负载，从而让计划在周用量上保持宽松。5 小时窗口用来摊平需求，周额度才是它要保护的东西。 Q: Pro 的豁免是永久的吗？ A: 没有公布截止日期。帖子的说法是接下来几个月对 Pro $100 与 Pro $200 订阅继续不启用 5 小时限制。把它当作随时可能变化的当前政策；有新帖改变它时本页会更新。 Q: Business 计划呢？ A: Business 在 2026-07-12 的移除里被点名，但 2026-08-25 的恢复帖没有提到它。本站不往任何方向推断；上表把它标为未确认，直到 OpenAI 明说。 Q: Pro 还有周限额吗？ A: 帮助中心说，视计划而定，限额可能按 5 小时窗口和周窗口计算。Pro 只是关掉了 5 小时窗口；帖子里没有任何一句说周限额被移除，而且 OpenAI 也没有公布周数字。 Sources: 1. Tibo Sottiaux on X, 2026-07-12 (5-hour limit removed for Plus, Business and Pro) — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 2. Tibo Sottiaux on X, 2026-07-14 (5h limit still off) — https://x.com/thsottiaux/status/2077114635308986427 (verified 2026-09-17) 3. Tibo Sottiaux on X, posted 2026-08-25 01:16 UTC ("Tomorrow we will bring back the 5h limit for Plus") — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) 4. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) ### Does reasoning effort count toward the Codex usage limit? URL: https://didreset.com/guides/does-reasoning-effort-count-codex Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 Yes. OpenAI's help center lists reasoning among the factors that drain Codex usage: 'Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools.' The GPT-6 Astra guide adds that 'Higher effort can use more of your allowance and does not always produce a better result.' Speed costs too: 'Fast mode applies a 2.5x multiplier to Astra's Standard rate.' Pick model, effort and speed before a long task, then check /status. Definition Reasoning is metered, not free. The pricing page says 'Model choice, context, reasoning, tool use, retrieval, and caching all affect usage, so prompt length alone isn't a reliable estimate.' The help center's plan article repeats it: 'Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools.' Effort is a dial you set separately from the model. The GPT-6 Astra guide describes lower effort as 'A useful starting point when you want a faster response or to make your allowance go further', and higher effort as useful for hard problems but 'Higher effort can use more of your allowance and does not always produce a better result.' It also warns that 'A reasoning level does not set a fixed amount of usage for a task or guarantee a better result.' Speed is a second multiplier. Fast mode 'provides faster responses and uses more of your included allowance', and the pricing page puts a number on it: 'Fast mode applies a 2.5x multiplier to Astra's Standard rate.' On 2026-07-10 Tibo Sottiaux wrote 'We made it too easy to use the highest-compute settings without making the impact on usage limits sufficiently clear.' What the official docs say - "Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools." — OpenAI Help Center: Using Codex with your ChatGPT plan (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) - "Higher effort can use more of your allowance and does not always produce a better result." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "A reasoning level does not set a fixed amount of usage for a task or guarantee a better result." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Model choice, context, reasoning, tool use, retrieval, and caching all affect usage, so prompt length alone isn't a reliable estimate." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Fast mode applies a 2.5x multiplier to Astra's Standard rate." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "We made it too easy to use the highest-compute settings without making the impact on usage limits sufficiently clear." — Tibo Sottiaux on X, 2026-07-10 (highest-compute settings and usage limits) (https://x.com/thsottiaux/status/2075641131002700120) How it is counted | Setting | Counted toward the 5-hour / weekly limit? | Source | | --- | --- | --- | | Reasoning effort: higher | Yes. "Higher effort can use more of your allowance" | [1] | | Reasoning effort: lower | Yes, but less: the starting point "to make your allowance go further" | [1] | | Fast mode | Yes. 2.5x multiplier to the Astra Standard rate | [2] | | Prompt length alone | Not a reliable estimate; model, context, reasoning, tool use, retrieval and caching all count | [2] | | Where the task runs, task complexity, tools | Yes. All listed as factors usage depends on | [3] | FAQ: Q: Does a lower reasoning level guarantee less usage? A: No. OpenAI says 'A reasoning level does not set a fixed amount of usage for a task or guarantee a better result.' Lower effort is the starting point 'to make your allowance go further', but the task itself still decides how much is used. Q: Does Fast mode use more of my Codex allowance? A: Yes. The help center says Fast mode 'provides faster responses and uses more of your included allowance', and the pricing page states 'Fast mode applies a 2.5x multiplier to Astra's Standard rate.' Q: Did OpenAI acknowledge that high-compute settings drained limits? A: On 2026-07-10 Tibo Sottiaux wrote: 'We made it too easy to use the highest-compute settings without making the impact on usage limits sufficiently clear.' Reasoning level and speed are now set separately from the model. Q: How do I see how much a task used? A: Type /status in an active Codex CLI session to see your remaining limits and reset time, or open Settings → Usage in the desktop app or on web. Your 5-hour and weekly windows are per account and are not the resets this board tracks. Sources: 1. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 3. OpenAI Help Center: Using Codex with your ChatGPT plan — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-07-10 (highest-compute settings and usage limits) — https://x.com/thsottiaux/status/2075641131002700120 (verified 2026-09-17) ### reasoning effort 算不算 Codex 的额度？ URL: https://didreset.com/zh/guides/does-reasoning-effort-count-codex Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 算。OpenAI 帮助中心把 reasoning 列为消耗 Codex 额度的因素之一，与模型、上下文、速度和工具并列。GPT-6 Astra 指南说更高的 effort 可能用掉更多额度，且不一定结果更好。定价页写明 Fast mode 按标准费率的 2.5 倍计。长任务前先选好模型、effort 和速度，再用 /status 看剩余。 定义 reasoning 是计量的，不是免费的。定价页说模型选择、上下文、reasoning、工具使用、检索和缓存都影响用量，所以光看提示词长度估不准。帮助中心的计划文章也重复了这一点：用量取决于模型、任务运行位置、任务复杂度、上下文、reasoning、速度和工具。 effort 是独立于模型的旋钮。GPT-6 Astra 指南把较低 effort 描述为想要更快回复或想让额度更耐用时的起点，较高 effort 适合难题，但可能用掉更多额度且不一定结果更好。它还提醒：reasoning 等级不会给一个任务设定固定用量，也不保证更好的结果。 速度是第二个乘数。Fast mode 回复更快、用掉更多包含额度，定价页给了数字：Fast mode 按 Astra 标准费率的 2.5 倍计。2026-07-10，Tibo Sottiaux 写道：我们让用户太容易用上最高算力的设置，却没有把它对额度的影响讲清楚。 官方文档怎么说 - "Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools." — OpenAI Help Center: Using Codex with your ChatGPT plan (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) - "Higher effort can use more of your allowance and does not always produce a better result." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "A reasoning level does not set a fixed amount of usage for a task or guarantee a better result." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Model choice, context, reasoning, tool use, retrieval, and caching all affect usage, so prompt length alone isn't a reliable estimate." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Fast mode applies a 2.5x multiplier to Astra's Standard rate." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "We made it too easy to use the highest-compute settings without making the impact on usage limits sufficiently clear." — Tibo Sottiaux on X, 2026-07-10 (highest-compute settings and usage limits) (https://x.com/thsottiaux/status/2075641131002700120) 怎么算 | 设置 | 算不算 5 小时 / 周额度？ | 来源 | | --- | --- | --- | | reasoning effort：高 | 算。更高的 effort 可能用掉更多额度 | [1] | | reasoning effort：低 | 算，但更少：这是让额度更耐用的起点 | [1] | | Fast mode | 算。按 Astra 标准费率的 2.5 倍 | [2] | | 只看提示词长度 | 估不准；模型、上下文、reasoning、工具、检索、缓存都算 | [2] | | 任务运行位置、任务复杂度、工具 | 算。都被列为用量取决的因素 | [3] | FAQ: Q: 把 reasoning 调低就一定省额度吗？ A: 不一定。OpenAI 说 reasoning 等级不会给一个任务设定固定用量，也不保证更好的结果。较低 effort 是让额度更耐用的起点，但用掉多少仍由任务本身决定。 Q: Fast mode 会多用 Codex 额度吗？ A: 会。帮助中心说 Fast mode 回复更快、用掉更多包含额度；定价页写明 Fast mode 按 Astra 标准费率的 2.5 倍计。 Q: OpenAI 承认过高算力设置把额度用得太快吗？ A: 2026-07-10，Tibo Sottiaux 写道：我们让用户太容易用上最高算力的设置，却没有把它对额度的影响讲清楚。现在 reasoning 等级和速度都独立于模型设置。 Q: 怎么看一个任务用了多少？ A: 在运行中的 Codex CLI 会话里输入 /status 可看到剩余限额与重置时间，桌面应用与网页版在 Settings → Usage。你的 5 小时与周窗口按账号各自计算，不是本站出发板追踪的全员重置。 Sources: 1. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 3. OpenAI Help Center: Using Codex with your ChatGPT plan — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-07-10 (highest-compute settings and usage limits) — https://x.com/thsottiaux/status/2075641131002700120 (verified 2026-09-17) ### Does extended thinking count toward the Claude usage limit? URL: https://didreset.com/guides/does-thinking-count-claude Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 Yes. Anthropic's help center says a higher effort level uses more tokens, 'so you'll reach your usage limits faster', and the Claude Code docs state 'Thinking tokens are billed as output tokens, and the default budget can be tens of thousands of tokens per request'. That output draws from one pool across claude.ai, Claude Code and Claude Desktop. Lower the effort with /effort or /model; thinking cannot be turned off on Fable 5.1 or Opus 5. Definition Thinking is output, and output is metered. The Claude Platform pricing page lists 'Tokens Claude uses while thinking (billed as output tokens)', and the Claude Code cost guide says 'Thinking tokens are billed as output tokens, and the default budget can be tens of thousands of tokens per request depending on the model.' Effort is the dial. The help center says 'Higher effort means more thorough responses, but they take longer and use more tokens, so you'll reach your usage limits faster.' Usage is also affected by which model you chat with and the effort level you selected, and 'your usage of all different Claude product surfaces (claude.ai, Claude Code, Claude Desktop) counts towards the same usage limit.' You can turn it down but not always off. In Claude Code, lower the effort level with /effort or in /model, or disable thinking in /config, but 'You can't turn off thinking on Fable models.' The help center says the same for the consumer apps: 'Thinking cannot be turned off in Claude when using Claude Fable 5.1 or Claude Opus 5.' What the official docs say - "Higher effort means more thorough responses, but they take longer and use more tokens, so you'll reach your usage limits faster." — Claude Help Center: Change the model, effort, and thinking settings (https://support.claude.com/en/articles/8664678-change-the-model-effort-and-thinking-settings) - "Thinking tokens are billed as output tokens, and the default budget can be tens of thousands of tokens per request depending on the model." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Tokens Claude uses while thinking (billed as output tokens)" — Claude Platform docs: Steering thinking and cost (https://platform.claude.com/docs/en/build-with-claude/thinking-steering-and-cost) - "your usage of all different Claude product surfaces (claude.ai, Claude Code, Claude Desktop) counts towards the same usage limit." — Claude Help Center: How do usage and length limits work? (https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work) - "Thinking cannot be turned off in Claude when using Claude Fable 5.1 or Claude Opus 5." — Claude Help Center: Change the model, effort, and thinking settings (https://support.claude.com/en/articles/8664678-change-the-model-effort-and-thinking-settings) - "You can't turn off thinking on Fable models." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) How it is counted | Setting | Counted toward the session / weekly limit? | Source | | --- | --- | --- | | Effort level: higher | Yes. More tokens, "so you'll reach your usage limits faster" | [1] | | Thinking tokens | Yes. Billed as output tokens; the default budget can be tens of thousands per request | [3] [4] | | Surface: claude.ai, Claude Code, Claude Desktop | All count toward the same usage limit | [2] | | Turning thinking off | Possible via /config in Claude Code, but not on Fable models; not on Fable 5.1 or Opus 5 in the apps | [1] [4] | FAQ: Q: Can I turn thinking off to save usage? A: Not on every model. The help center says 'Thinking cannot be turned off in Claude when using Claude Fable 5.1 or Claude Opus 5', and the Claude Code docs say 'You can't turn off thinking on Fable models.' On other models you can disable it in /config or lower the effort with /effort. Q: Do claude.ai and Claude Code share the same limit? A: Yes. Anthropic says 'your usage of all different Claude product surfaces (claude.ai, Claude Code, Claude Desktop) counts towards the same usage limit.' Thinking tokens spent in one surface come out of the pool the others use. Q: Are thinking tokens cheaper than normal output? A: No. The Claude Platform pricing page lists 'Tokens Claude uses while thinking (billed as output tokens)', and the Claude Code docs repeat that thinking tokens are billed as output tokens with a default budget that 'can be tens of thousands of tokens per request'. Q: How do I lower the effort level in Claude Code? A: The Claude Code cost guide says you can lower the effort level with /effort or in /model, or disable thinking in /config where the model allows it. Run /usage to see the session and weekly bars afterwards. Sources: 1. Claude Help Center: Change the model, effort, and thinking settings — https://support.claude.com/en/articles/8664678-change-the-model-effort-and-thinking-settings (verified 2026-09-17) 2. Claude Help Center: How do usage and length limits work? — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) 3. Claude Platform docs: Steering thinking and cost — https://platform.claude.com/docs/en/build-with-claude/thinking-steering-and-cost (verified 2026-09-17) 4. Claude Code docs: Manage costs effectively — https://code.claude.com/docs/en/costs (verified 2026-09-17) ### extended thinking 算不算 Claude 的额度？ URL: https://didreset.com/zh/guides/does-thinking-count-claude Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 算。Anthropic 帮助中心说更高的 effort 会用更多 token，更快碰到限额；Claude Code 文档写明 thinking token 按输出 token 计费，预算可达每次数万。claude.ai、Claude Code、Desktop 共用一个池子。用 /effort 或 /model 调低；Fable 5.1 与 Opus 5 关不掉。 定义 thinking 就是输出，输出是计量的。Claude Platform 定价页把模型思考时用的 token 列为按输出 token 计费；Claude Code 成本指南说 thinking token 按输出 token 计费，默认预算视模型而定，每次请求可达数万 token。 effort 是旋钮。帮助中心说更高的 effort 意味着更周全的回答，但耗时更长、用更多 token，所以你会更快碰到限额。用量还受你在用的模型和所选 effort 等级影响，而且所有 Claude 产品入口（claude.ai、Claude Code、Claude Desktop）的用量都计入同一个限额。 可以调低，但不一定能关。在 Claude Code 里可用 /effort 或在 /model 里调低 effort，也可以在 /config 里关闭 thinking，但 Fable 系列模型无法关闭。帮助中心对消费级应用的说法一样：使用 Claude Fable 5.1 或 Claude Opus 5 时无法关闭 thinking。 官方文档怎么说 - "Higher effort means more thorough responses, but they take longer and use more tokens, so you'll reach your usage limits faster." — Claude Help Center: Change the model, effort, and thinking settings (https://support.claude.com/en/articles/8664678-change-the-model-effort-and-thinking-settings) - "Thinking tokens are billed as output tokens, and the default budget can be tens of thousands of tokens per request depending on the model." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Tokens Claude uses while thinking (billed as output tokens)" — Claude Platform docs: Steering thinking and cost (https://platform.claude.com/docs/en/build-with-claude/thinking-steering-and-cost) - "your usage of all different Claude product surfaces (claude.ai, Claude Code, Claude Desktop) counts towards the same usage limit." — Claude Help Center: How do usage and length limits work? (https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work) - "Thinking cannot be turned off in Claude when using Claude Fable 5.1 or Claude Opus 5." — Claude Help Center: Change the model, effort, and thinking settings (https://support.claude.com/en/articles/8664678-change-the-model-effort-and-thinking-settings) - "You can't turn off thinking on Fable models." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) 怎么算 | 设置 | 算不算会话 / 周额度？ | 来源 | | --- | --- | --- | | effort 等级：高 | 算。用更多 token，更快碰到限额 | [1] | | thinking token | 算。按输出 token 计费；默认预算每次请求可达数万 | [3] [4] | | 入口：claude.ai、Claude Code、Claude Desktop | 都计入同一个限额 | [2] | | 关闭 thinking | Claude Code 可在 /config 里关，但 Fable 系列不行；应用里 Fable 5.1 与 Opus 5 也不行 | [1] [4] | FAQ: Q: 能不能关掉 thinking 来省额度？ A: 不是每个模型都行。帮助中心说使用 Claude Fable 5.1 或 Claude Opus 5 时无法关闭 thinking，Claude Code 文档也说 Fable 系列模型关不掉。其他模型可以在 /config 里关闭，或用 /effort 调低。 Q: claude.ai 和 Claude Code 共用一个限额吗？ A: 共用。Anthropic 说所有 Claude 产品入口（claude.ai、Claude Code、Claude Desktop）的用量都计入同一个限额。在一个入口花掉的 thinking token，其他入口就少了。 Q: thinking token 比普通输出便宜吗？ A: 不便宜。Claude Platform 定价页把模型思考时用的 token 列为按输出 token 计费；Claude Code 文档也重复说 thinking token 按输出 token 计费，默认预算每次请求可达数万 token。 Q: 在 Claude Code 里怎么调低 effort？ A: Claude Code 成本指南说可以用 /effort 或在 /model 里调低 effort，模型允许时也可以在 /config 里关闭 thinking。之后用 /usage 查看会话与周额度条。 Sources: 1. Claude Help Center: Change the model, effort, and thinking settings — https://support.claude.com/en/articles/8664678-change-the-model-effort-and-thinking-settings (verified 2026-09-17) 2. Claude Help Center: How do usage and length limits work? — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) 3. Claude Platform docs: Steering thinking and cost — https://platform.claude.com/docs/en/build-with-claude/thinking-steering-and-cost (verified 2026-09-17) 4. Claude Code docs: Manage costs effectively — https://code.claude.com/docs/en/costs (verified 2026-09-17) ### Do subagents count toward the Claude Code usage limit? URL: https://didreset.com/guides/do-subagents-count-claude-code Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 Yes. The Claude Code docs say the /usage plan breakdown shows 'what counts against your plan limits', with usage 'attributed to skills, subagents, plugins, and individual MCP servers'. Each subagent runs in its own context window, so its tokens come on top of the main conversation, and 'Agent teams use approximately 7x more tokens than standard sessions when teammates run in plan mode.' To spend less, set model: haiku for simple subagent tasks. Definition Subagents are metered against your plan. The cost guide says that on a Pro, Max, Team or Enterprise plan '/usage also shows a breakdown of what counts against your plan limits: Attribution: recent usage attributed to skills, subagents, plugins, and individual MCP servers'. Each one is a separate context. The subagent docs say 'Each subagent runs in its own context window with a custom system prompt, specific tool access, and independent permissions', so a subagent's tokens are added to, not shared with, the main conversation. For agent teams, 'Token usage scales with the number of active teammates and how long each one runs', and 'Agent teams use approximately 7x more tokens than standard sessions when teammates run in plan mode.' They can still save main-context tokens. Verbose tool output stays in the subagent's context while only a summary returns to your main conversation, which keeps later turns cheaper. The cost guide's lever for the subagent side: 'For simple subagent tasks, specify model: haiku in your subagent configuration.' What the official docs say - "/usage also shows a breakdown of what counts against your plan limits: Attribution: recent usage attributed to skills, subagents, plugins, and individual MCP servers" — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Each subagent runs in its own context window with a custom system prompt, specific tool access, and independent permissions" — Claude Code docs: Create custom subagents (https://code.claude.com/docs/en/sub-agents) - "Token usage scales with the number of active teammates and how long each one runs" — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Agent teams use approximately 7x more tokens than standard sessions when teammates run in plan mode." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "For simple subagent tasks, specify model: haiku in your subagent configuration." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) How it is counted | Feature | Counted toward the session / weekly limit? | Source | | --- | --- | --- | | Subagents | Yes. Listed in the /usage attribution breakdown of what counts against plan limits | [1] | | Subagent context | Own context window per subagent, added on top of the main conversation | [2] | | Agent teams | Yes. Approximately 7x more tokens than standard sessions when teammates run in plan mode | [1] | | Skills, plugins, MCP servers | Yes. Attributed in the same breakdown | [1] | | Cheaper subagents | Set model: haiku for simple subagent tasks | [1] | FAQ: Q: Do subagents save usage or cost more? A: Both can be true. Their tokens count against your plan, and each runs in its own context window. But verbose output stays in the subagent's context while only a summary returns to your main conversation, so the main thread stays smaller for later turns. Q: How much more do agent teams use? A: The cost guide says 'Agent teams use approximately 7x more tokens than standard sessions when teammates run in plan mode', and that 'Token usage scales with the number of active teammates and how long each one runs'. Q: Where do I see what my subagents used? A: Run /usage in Claude Code. On a Pro, Max, Team or Enterprise plan it shows 'a breakdown of what counts against your plan limits', including recent usage attributed to skills, subagents, plugins and individual MCP servers. Q: How do I make subagents cheaper? A: The cost guide says: 'For simple subagent tasks, specify model: haiku in your subagent configuration.' Keep heavier models for the subagents that need them. Sources: 1. Claude Code docs: Manage costs effectively — https://code.claude.com/docs/en/costs (verified 2026-09-17) 2. Claude Code docs: Create custom subagents — https://code.claude.com/docs/en/sub-agents (verified 2026-09-17) ### subagent 算不算 Claude Code 的额度？ URL: https://didreset.com/zh/guides/do-subagents-count-claude-code Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 算。Claude Code 文档说 /usage 的计划用量分解会把计入计划限额的用量归因到 skill、subagent、插件和 MCP 服务器。每个 subagent 有自己的上下文窗口，token 叠加在主对话之上；plan mode 下的 agent team 约用 7 倍 token。简单任务设成 model: haiku 可省额度。 定义 subagent 按计划计量。成本指南说，在 Pro、Max、Team 或 Enterprise 计划上，/usage 还会显示计入计划限额的用量分解，其中的归因一项列出最近归到 skill、subagent、插件和各个 MCP 服务器的用量。 每个 subagent 是独立的上下文。subagent 文档说每个 subagent 在自己的上下文窗口里运行，有自定义系统提示、特定工具权限和独立的权限设置，所以它的 token 是加在主对话之上，不是和主对话共用。对于 agent team，token 用量随活跃队友数量和每个队友的运行时长增长；当队友在 plan mode 下运行时，agent team 用的 token 约为普通会话的 7 倍。 它们仍然能省主上下文的 token。冗长的工具输出留在 subagent 的上下文里，只有摘要回到主对话，后续回合因此更便宜。成本指南给 subagent 这一侧的省钱办法是：简单的 subagent 任务在配置里指定 model: haiku。 官方文档怎么说 - "/usage also shows a breakdown of what counts against your plan limits: Attribution: recent usage attributed to skills, subagents, plugins, and individual MCP servers" — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Each subagent runs in its own context window with a custom system prompt, specific tool access, and independent permissions" — Claude Code docs: Create custom subagents (https://code.claude.com/docs/en/sub-agents) - "Token usage scales with the number of active teammates and how long each one runs" — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Agent teams use approximately 7x more tokens than standard sessions when teammates run in plan mode." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "For simple subagent tasks, specify model: haiku in your subagent configuration." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) 怎么算 | 功能 | 算不算会话 / 周额度？ | 来源 | | --- | --- | --- | | subagent | 算。列在 /usage 里计入计划限额的归因分解中 | [1] | | subagent 的上下文 | 每个 subagent 各有自己的上下文窗口，叠加在主对话之上 | [2] | | agent team | 算。队友在 plan mode 下运行时约为普通会话的 7 倍 token | [1] | | skill、插件、MCP 服务器 | 算。在同一个分解里归因 | [1] | | 更便宜的 subagent | 简单任务设成 model: haiku | [1] | FAQ: Q: subagent 是省额度还是更费额度？ A: 两者都可能。它们的 token 计入你的计划，而且每个都在自己的上下文窗口里运行。但冗长的输出留在 subagent 的上下文里，只有摘要回到主对话，所以主线程在后续回合里更小。 Q: agent team 会多用多少？ A: 成本指南说，当队友在 plan mode 下运行时，agent team 用的 token 约为普通会话的 7 倍；token 用量随活跃队友数量和每个队友的运行时长增长。 Q: 在哪里看 subagent 用了多少？ A: 在 Claude Code 里运行 /usage。Pro、Max、Team 或 Enterprise 计划上它会显示计入计划限额的用量分解，包括最近归到 skill、subagent、插件和各个 MCP 服务器的用量。 Q: 怎么让 subagent 更省？ A: 成本指南说：简单的 subagent 任务在 subagent 配置里指定 model: haiku。把更重的模型留给真正需要的 subagent。 Sources: 1. Claude Code docs: Manage costs effectively — https://code.claude.com/docs/en/costs (verified 2026-09-17) 2. Claude Code docs: Create custom subagents — https://code.claude.com/docs/en/sub-agents (verified 2026-09-17) ### Do prompt cache reads count toward the Claude Code and Codex usage limits? URL: https://didreset.com/guides/does-prompt-cache-count Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 Yes. Cache reads still count, at a reduced rate. Claude Code re-reads history at the cached token rate, so 'a one-line question in a session that has been open all day still draws usage for the whole conversation'; cache reads cost 0.1 times base input, and a break longer than the one-hour subscription cache lifetime triggers a full re-read. OpenAI lists caching among the factors that affect Codex usage, with cached input at one tenth of the input rate. Definition On Claude, a cache hit is cheaper input, not free input. The Platform docs say 'Cache read tokens are 0.1 times the base input tokens price' (0.025x on Claude Fable 5.1 and Mythos 5.1), while cache writes cost 1.25x base input for a 5-minute cache and 2x for a 1-hour cache. Claude Code applies that to your whole conversation. 'With prompt caching, Claude Code re-reads that history at the cached token rate', which is why 'a one-line question in a session that has been open all day still draws usage for the whole conversation'. Misses are worse: 'Cache misses: your first message after a break longer than the cache lifetime misses the cache and reprocesses your full context', and 'The lifetime is an hour on a subscription'. The /usage breakdown flags long context or cache misses when they account for 10% or more of recent usage. Codex says the same in fewer words: 'Model choice, context, reasoning, tool use, retrieval, and caching all affect usage'. Its credit price list bills cached input at one tenth of input (for GPT-6 Astra, 25 credits per million cached tokens against 250 for input). On 2026-05-23 Tibo Sottiaux traced a run of fast-draining limits to an optimization 'that had an impact on cache hit rates', which was rolled back. What the official docs say - "Cache read tokens are 0.1 times the base input tokens price" — Claude Platform docs: Prompt caching (https://platform.claude.com/docs/en/build-with-claude/prompt-caching) - "a one-line question in a session that has been open all day still draws usage for the whole conversation" — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Cache misses: your first message after a break longer than the cache lifetime misses the cache and reprocesses your full context." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "The lifetime is an hour on a subscription" — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Model choice, context, reasoning, tool use, retrieval, and caching all affect usage" — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "we root caused it to an optimization that we rolled back that had an impact on cache hit rates" — Tibo Sottiaux on X, 2026-05-23 (cache hit rate optimization rolled back) (https://x.com/thsottiaux/status/2058280452851638313) How it is counted | Provider | Item | Counted? | Source | | --- | --- | --- | --- | | Claude Code | Cache read | Yes. 0.1x base input price (0.025x on Fable 5.1 and Mythos 5.1) | [1] | | Claude Code | Cache write | Yes. 1.25x base input for 5-minute caches, 2x for 1-hour caches | [1] | | Claude Code | Cache miss after a break | Yes. Full context reprocessed; the lifetime is an hour on a subscription | [2] | | Codex | Caching | Yes. Listed among the factors that all affect usage; cached input is one tenth of input on the credit price list | [3] | FAQ: Q: Why does a one-line question in a long Claude Code session use so much? A: Because the whole history is re-read. The Claude Code docs say that with prompt caching it 're-reads that history at the cached token rate, so a one-line question in a session that has been open all day still draws usage for the whole conversation'. Q: How long does the Claude Code cache last on a subscription? A: An hour. The docs say 'The lifetime is an hour on a subscription', and that 'your first message after a break longer than the cache lifetime misses the cache and reprocesses your full context'. Starting a fresh session after a long break avoids paying for a stale history. Q: Does Codex charge for cached input? A: Yes, at a reduced rate. The pricing page lists caching among the factors that 'all affect usage', and its credit price list bills cached input at one tenth of the input rate (25 versus 250 credits per million tokens for GPT-6 Astra). Q: Has caching ever made limits drain faster? A: On 2026-05-23 Tibo Sottiaux wrote that Codex had 'root caused it to an optimization that we rolled back that had an impact on cache hit rates'. Lower hit rates mean more full-price input, which is what users saw as limits draining faster. Sources: 1. Claude Platform docs: Prompt caching — https://platform.claude.com/docs/en/build-with-claude/prompt-caching (verified 2026-09-17) 2. Claude Code docs: Manage costs effectively — https://code.claude.com/docs/en/costs (verified 2026-09-17) 3. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-05-23 (cache hit rate optimization rolled back) — https://x.com/thsottiaux/status/2058280452851638313 (verified 2026-09-17) ### prompt cache 缓存读取算不算 Claude Code 和 Codex 的额度？ URL: https://didreset.com/zh/guides/does-prompt-cache-count Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 算。缓存读取照样计入，只是打折。Claude Code 按缓存 token 费率重读历史，所以在开了一整天的会话里问一句话，仍会为整段对话扣额度；缓存读取按基础输入价的 0.1 倍计，停顿超过订阅版一小时的缓存有效期就会整段重读。OpenAI 把缓存列为影响 Codex 用量的因素之一，缓存输入按输入价的十分之一计。 定义 在 Claude 上，缓存命中是更便宜的输入，不是免费输入。Platform 文档说缓存读取 token 按基础输入 token 价格的 0.1 倍计（Claude Fable 5.1 与 Mythos 5.1 为 0.025 倍），缓存写入则是 5 分钟缓存 1.25 倍、1 小时缓存 2 倍。 Claude Code 把这套规则套在整段对话上。开启 prompt caching 后它按缓存 token 费率重读历史，所以在开了一整天的会话里问一句话仍然会为整段对话扣额度。未命中更糟：停顿超过缓存有效期后的第一条消息会未命中缓存并重新处理全部上下文，而订阅版的有效期是一小时。/usage 的分解会在长上下文或缓存未命中占最近用量 10% 以上时标出来。 Codex 用更少的字说了同样的话：模型选择、上下文、reasoning、工具使用、检索和缓存都影响用量。它的积分价目表把缓存输入按输入的十分之一计（GPT-6 Astra 每百万缓存 token 25 积分，输入为 250）。2026-05-23，Tibo Sottiaux 把一轮额度消耗过快追溯到一个影响缓存命中率的优化，并已回滚。 官方文档怎么说 - "Cache read tokens are 0.1 times the base input tokens price" — Claude Platform docs: Prompt caching (https://platform.claude.com/docs/en/build-with-claude/prompt-caching) - "a one-line question in a session that has been open all day still draws usage for the whole conversation" — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Cache misses: your first message after a break longer than the cache lifetime misses the cache and reprocesses your full context." — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "The lifetime is an hour on a subscription" — Claude Code docs: Manage costs effectively (https://code.claude.com/docs/en/costs) - "Model choice, context, reasoning, tool use, retrieval, and caching all affect usage" — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "we root caused it to an optimization that we rolled back that had an impact on cache hit rates" — Tibo Sottiaux on X, 2026-05-23 (cache hit rate optimization rolled back) (https://x.com/thsottiaux/status/2058280452851638313) 怎么算 | 厂商 | 项目 | 算不算？ | 来源 | | --- | --- | --- | --- | | Claude Code | 缓存读取 | 算。基础输入价的 0.1 倍（Fable 5.1 与 Mythos 5.1 为 0.025 倍） | [1] | | Claude Code | 缓存写入 | 算。5 分钟缓存 1.25 倍、1 小时缓存 2 倍基础输入价 | [1] | | Claude Code | 停顿后的缓存未命中 | 算。重新处理全部上下文；订阅版有效期一小时 | [2] | | Codex | 缓存 | 算。列在影响用量的因素里；积分价目表上缓存输入为输入的十分之一 | [3] | FAQ: Q: 为什么在很长的 Claude Code 会话里问一句话也很费额度？ A: 因为整段历史都被重读了。Claude Code 文档说开启 prompt caching 后它按缓存 token 费率重读历史，所以在开了一整天的会话里问一句话，仍然会为整段对话扣额度。 Q: 订阅版 Claude Code 的缓存能保持多久？ A: 一小时。文档说订阅版的有效期是一小时，停顿超过缓存有效期后的第一条消息会未命中缓存并重新处理全部上下文。长时间离开后新开会话，可以避免为一段过期历史付费。 Q: Codex 会为缓存输入扣额度吗？ A: 会，按折扣计。定价页把缓存列在影响用量的因素里，积分价目表把缓存输入按输入价的十分之一计（GPT-6 Astra 每百万 token 25 积分，输入为 250）。 Q: 缓存出过让额度掉得更快的问题吗？ A: 2026-05-23，Tibo Sottiaux 写道 Codex 已把问题追溯到一个影响缓存命中率的优化并回滚。命中率降低意味着更多全价输入，用户看到的就是额度掉得更快。 Sources: 1. Claude Platform docs: Prompt caching — https://platform.claude.com/docs/en/build-with-claude/prompt-caching (verified 2026-09-17) 2. Claude Code docs: Manage costs effectively — https://code.claude.com/docs/en/costs (verified 2026-09-17) 3. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-05-23 (cache hit rate optimization rolled back) — https://x.com/thsottiaux/status/2058280452851638313 (verified 2026-09-17) ### Does web search count toward the Claude and Codex usage limits? URL: https://didreset.com/guides/does-web-search-count Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 Yes. Anthropic states it outright: 'Usage of web search and web fetch counts toward your usage limits', and warns that fetching a long article uses far more than a search query. OpenAI does not name web search but says Codex usage depends on 'tools' alongside model, context and reasoning. xAI has not published a per-feature rule for Grok; it meters one weekly pool where 'Different products cost different amounts depending on how much compute that product requires'. Definition Claude is explicit. The help center says 'Usage of web search and web fetch counts toward your usage limits.' Fetching is the expensive half: asking Claude to summarize a 10,000-word article via web fetch 'will use substantially more of your context window than a regular web search query'. Codex counts tools without naming them. The help center says 'Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools.' Web search is a tool, so it draws on the same allowance, but OpenAI publishes no per-tool rate. xAI has not published a per-feature rule. The Grok FAQ describes one weekly pool across products and says 'Different products cost different amounts depending on how much compute that product requires.' This site does not turn that into a web-search figure. What the official docs say - "Usage of web search and web fetch counts toward your usage limits." — Claude Help Center: Enable and use web search (https://support.claude.com/en/articles/10684626-enable-and-use-web-search) - "will use substantially more of your context window than a regular web search query" — Claude Help Center: Enable and use web search (https://support.claude.com/en/articles/10684626-enable-and-use-web-search) - "Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools." — OpenAI Help Center: Using Codex with your ChatGPT plan (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) - "Different products cost different amounts depending on how much compute that product requires." — xAI docs: FAQ – Grok website and apps (https://docs.x.ai/grok/faq) How it is counted | Provider | Feature | Counted? | Source | | --- | --- | --- | --- | | Claude (claude.ai, Claude Code) | Web search | Yes. Counts toward your usage limits | [1] | | Claude | Web fetch of a long page | Yes, and substantially more context than a regular web search query | [1] | | Codex | Tools (web search not named separately) | Yes. Usage depends on "tools" among other factors; no per-tool rate published | [2] | | Grok | Any product | Not published per feature. One weekly pool; different products cost different amounts | [3] | FAQ: Q: Does web fetch cost more than web search on Claude? A: Yes. The help center says summarizing a 10,000-word article via web fetch 'will use substantially more of your context window than a regular web search query'. Both count toward your usage limits. Q: Does OpenAI say web search counts for Codex? A: Not by name. The help center says 'Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools.' Web search is one of those tools; OpenAI has not published a separate rate for it. Q: Does Grok charge web search separately? A: xAI has not published a per-feature rule. The Grok FAQ says 'Different products cost different amounts depending on how much compute that product requires', and that usage is drawn from one weekly pool shown in the Usage tab in Settings. Q: Where do I see the effect on my own limit? A: Run /usage in Claude Code, /status in the Codex CLI, or open the Usage tab in Grok Settings. These are your per-account windows; the resets on the DidReset board are account-wide announcements and do not change how tools are counted. Sources: 1. Claude Help Center: Enable and use web search — https://support.claude.com/en/articles/10684626-enable-and-use-web-search (verified 2026-09-17) 2. OpenAI Help Center: Using Codex with your ChatGPT plan — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 3. xAI docs: FAQ – Grok website and apps — https://docs.x.ai/grok/faq (verified 2026-09-17) ### 联网搜索算不算 Claude 和 Codex 的额度？ URL: https://didreset.com/zh/guides/does-web-search-count Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 算。Anthropic 说得很直接：web search 与 web fetch 的用量计入你的额度，并提醒抓取一篇长文章比一次搜索费得多。OpenAI 没有点名 web search，但说 Codex 用量取决于工具等因素。xAI 没有按功能公布 Grok 的规则，只说从一个周用量池计量，不同产品按所需算力收取不同用量。 定义 Claude 是明确的。帮助中心说 web search 与 web fetch 的用量计入你的额度。抓取是更贵的那一半：让 Claude 通过 web fetch 总结一篇一万词的文章，会比一次普通的 web search 查询多用得多的上下文窗口。 Codex 算工具但没点名。帮助中心说用量取决于模型、任务运行位置、任务复杂度、上下文、reasoning、速度和工具。联网搜索是工具之一，所以它从同一份额度里扣，但 OpenAI 没有公布按工具的费率。 xAI 没有按功能公布规则。Grok FAQ 描述的是跨产品的一个周用量池，并说不同产品按所需算力收取不同用量。本站不把它换算成联网搜索的数字。 官方文档怎么说 - "Usage of web search and web fetch counts toward your usage limits." — Claude Help Center: Enable and use web search (https://support.claude.com/en/articles/10684626-enable-and-use-web-search) - "will use substantially more of your context window than a regular web search query" — Claude Help Center: Enable and use web search (https://support.claude.com/en/articles/10684626-enable-and-use-web-search) - "Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools." — OpenAI Help Center: Using Codex with your ChatGPT plan (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) - "Different products cost different amounts depending on how much compute that product requires." — xAI docs: FAQ – Grok website and apps (https://docs.x.ai/grok/faq) 怎么算 | 厂商 | 功能 | 算不算？ | 来源 | | --- | --- | --- | --- | | Claude（claude.ai、Claude Code） | web search | 算。计入你的额度 | [1] | | Claude | 抓取长网页（web fetch） | 算，而且比一次普通搜索多用得多的上下文 | [1] | | Codex | 工具（未单独点名 web search） | 算。用量取决于工具等因素；没有公布按工具的费率 | [2] | | Grok | 任何产品 | 未按功能公布。一个周用量池；不同产品收取不同用量 | [3] | FAQ: Q: 在 Claude 上 web fetch 比 web search 更费吗？ A: 更费。帮助中心说通过 web fetch 总结一篇一万词的文章，会比一次普通的 web search 查询多用得多的上下文窗口。两者都计入你的额度。 Q: OpenAI 说过 Codex 的联网搜索算额度吗？ A: 没有点名。帮助中心说用量取决于模型、任务运行位置、任务复杂度、上下文、reasoning、速度和工具。联网搜索是这些工具之一；OpenAI 没有为它单独公布费率。 Q: Grok 的联网搜索单独收费吗？ A: xAI 没有按功能公布规则。Grok FAQ 说不同产品按所需算力收取不同用量，用量从设置里 Usage 标签页显示的一个周用量池里扣。 Q: 在哪里看它对我自己额度的影响？ A: 在 Claude Code 里运行 /usage，在 Codex CLI 里输入 /status，或打开 Grok 设置里的 Usage 标签页。这些都是你账号自己的窗口；DidReset 出发板上的重置是全员公告，不会改变工具的计量方式。 Sources: 1. Claude Help Center: Enable and use web search — https://support.claude.com/en/articles/10684626-enable-and-use-web-search (verified 2026-09-17) 2. OpenAI Help Center: Using Codex with your ChatGPT plan — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 3. xAI docs: FAQ – Grok website and apps — https://docs.x.ai/grok/faq (verified 2026-09-17) ### Does a banked reset change the Codex reset countdown? URL: https://didreset.com/guides/reset-card-countdown Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 Yes. OpenAI's help center says 'Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date. Check Settings → Usage for your updated reset time.' A purchased instant reset also brings the weekly allowance forward: the new weekly period starts with your first request after the reset, and the next automatic weekly reset is seven days later. Global resets apply on their own and never appear as a saved reset. Definition A banked reset is applied by you, and the moment it lands both windows start over. The help center says it 'refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date', and warns that 'Your usage dashboard may take a short time to update'; if the percentage does not move, refresh or reopen Settings → Usage. Tibo Sottiaux confirmed on 2026-07-13 that applying one 'will replenish the weekly usage for you'. A purchased instant reset moves the countdown the same way. The help center says 'Purchasing a reset brings your weekly allowance forward. Your new weekly period starts with your first Work or Codex request after the reset', and the next automatic weekly reset is seven days after that request. An automatic or global reset, the kind this board tracks, is different: it 'refreshes the eligible usage limits specified in the announcement', you do not need to apply it, and 'it won't appear as a saved reset in Settings'. After any reset the normal caps still apply: 'you can reach the five-hour limit while you still have weekly usage remaining'. What the official docs say - "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Check Settings → Usage for your updated reset time." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Your usage dashboard may take a short time to update." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Purchasing a reset brings your weekly allowance forward." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "An automatic or global reset refreshes the eligible usage limits specified in the announcement." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "it won't appear as a saved reset in Settings" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you." — Tibo Sottiaux on X, 2026-07-13 (banked reset replenishes weekly usage) (https://x.com/thsottiaux/status/2076735790567338203) How it is counted | Kind of reset | What happens to your countdown | Source | | --- | --- | --- | | Full banked reset (you apply it) | Both the 5-hour and weekly windows refresh; your weekly reset date changes | [1] | | Purchased instant reset | Weekly allowance brought forward; a new weekly period starts with your first request and the next automatic reset is seven days after it | [2] | | Automatic / global reset (what this board tracks) | Refreshes the limits named in the announcement; nothing to apply; not shown as a saved reset | [2] | | Dashboard after any reset | May take a short time to update; refresh or reopen Settings → Usage | [1] | | Caps after a reset | Unchanged: you can still reach the 5-hour limit with weekly usage remaining | [2] | FAQ: Q: Do I need to do anything when OpenAI announces a global reset? A: No. The help center says an automatic or global reset 'refreshes the eligible usage limits specified in the announcement', that you do not need to apply it yourself, and that it 'won't appear as a saved reset in Settings'. Only banked resets are applied by you. Q: Why does my dashboard still show the old percentage after a reset? A: OpenAI says 'Your usage dashboard may take a short time to update.' If the number has not changed, refresh the page or reopen Settings → Usage; in the CLI, /status reads the current windows. Q: Does a banked reset refill the weekly allowance too, or only the 5-hour window? A: Both. The help center names the '5-hour and weekly Codex usage windows', and Tibo Sottiaux wrote on 2026-07-13 that applying the reset 'will replenish the weekly usage for you'. Your weekly reset date moves accordingly. Q: After a reset, can I still hit the 5-hour limit? A: Yes. The help center says that after a reset 'you can reach the five-hour limit while you still have weekly usage remaining'. A reset refills the windows; it does not remove them. Sources: 1. OpenAI Help Center: How banked Codex resets work — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Tibo Sottiaux on X, 2026-07-13 (banked reset replenishes weekly usage) — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) ### 用了重置卡，Codex 的重置倒计时会变吗？ URL: https://didreset.com/zh/guides/reset-card-countdown Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 会。OpenAI 帮助中心写明：用掉一张完整重置卡会刷新你的 5 小时与周两个 Codex 窗口，并改变你的周重置日期，请到 Settings → Usage 查看更新后的重置时间。购买的即时重置同样会把周额度提前：新的周周期从重置后你的第一次请求开始，下一次自动周重置在其 7 天之后。全员重置自动生效，从不显示为已保存的重置。 定义 重置卡由你自己兑换，落地那一刻两个窗口都从头开始。帮助中心说它会刷新 5 小时与周两个 Codex 窗口并改变周重置日期，同时提醒用量面板可能要过一会儿才更新；百分比没动就刷新页面或重新打开 Settings → Usage。Tibo Sottiaux 在 2026-07-13 确认，兑换后会把周额度补回来。 购买的即时重置以同样方式移动倒计时。帮助中心说购买重置会把周额度提前，新的周周期从重置后你在 Work 或 Codex 的第一次请求开始，下一次自动周重置在那次请求之后的第 7 天。 自动重置或全员重置——也就是本站出发板追踪的那种——不一样：它刷新公告里指定的那些限额，你不需要自己去兑换，也不会在设置里显示为已保存的重置。任何重置之后正常上限照旧：周额度还有剩余时，你仍可能碰到 5 小时上限。 官方文档怎么说 - "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Check Settings → Usage for your updated reset time." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Your usage dashboard may take a short time to update." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Purchasing a reset brings your weekly allowance forward." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "An automatic or global reset refreshes the eligible usage limits specified in the announcement." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "it won't appear as a saved reset in Settings" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you." — Tibo Sottiaux on X, 2026-07-13 (banked reset replenishes weekly usage) (https://x.com/thsottiaux/status/2076735790567338203) 怎么算 | 重置类型 | 倒计时会怎样 | 来源 | | --- | --- | --- | | 完整重置卡（你自己兑换） | 5 小时与周两个窗口都刷新；周重置日期改变 | [1] | | 购买的即时重置 | 周额度提前；新的周周期从你的第一次请求开始，下一次自动重置在其 7 天之后 | [2] | | 自动 / 全员重置（本站出发板追踪的那种） | 刷新公告里点名的限额；不用兑换；不显示为已保存的重置 | [2] | | 任何重置之后的面板 | 可能要过一会儿才更新；刷新或重新打开 Settings → Usage | [1] | | 重置之后的上限 | 不变：周额度还有剩余时仍可能碰到 5 小时上限 | [2] | FAQ: Q: OpenAI 公告全员重置时，我需要做什么吗？ A: 不需要。帮助中心说自动或全员重置会刷新公告里指定的限额，不用你自己兑换，也不会在设置里显示为已保存的重置。只有重置卡才需要你自己动手。 Q: 重置之后面板为什么还显示旧的百分比？ A: OpenAI 说用量面板可能要过一会儿才更新。数字没变就刷新页面或重新打开 Settings → Usage；在 CLI 里输入 /status 读到的是当前窗口。 Q: 重置卡只补 5 小时窗口，还是连周额度也补？ A: 两个都补。帮助中心点名的是 5 小时与周两个 Codex 窗口，Tibo Sottiaux 在 2026-07-13 也写道兑换后会把周额度补回来。你的周重置日期会随之移动。 Q: 重置之后还会碰到 5 小时上限吗？ A: 会。帮助中心说重置之后，周额度还有剩余时你仍可能碰到 5 小时上限。重置是把窗口填满，不是把窗口拿掉。 Sources: 1. OpenAI Help Center: How banked Codex resets work — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Tibo Sottiaux on X, 2026-07-13 (banked reset replenishes weekly usage) — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) ### Do Codex banked resets expire? URL: https://didreset.com/guides/do-banked-resets-expire Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 Yes. The Codex pricing page states banked rate-limit resets 'are usable for 30 days after they're granted' (that sentence sits in the June 2026 referral-promotion terms), and the help center says expiration can 'vary by offer, plan, workspace, and region', so check the date shown in your account: 'Once an unused reset expires, it cannot be restored or reissued.' Banked resets are not cash, API credits or transferable balances, and Tibo Sottiaux noted some users had stacked up to three. Definition The only published number is 30 days. The pricing page says 'Banked rate-limit resets are usable for 30 days after they're granted.' That line is part of the June 11–24, 2026 referral-promotion terms, so it describes those grants rather than every offer. The help center makes the general rule per offer: 'Eligibility, affected usage limits, delivery timing, and expiration vary by offer, plan, workspace, and region.' It tells you to 'Check the expiration shown in your account or offer', and is blunt about what happens afterwards: 'Once an unused reset expires, it cannot be restored or reissued.' Banked resets are promotional, not money: 'Banked resets are not cash, API credits, or transferable credit balances.' You can hold more than one; on 2026-06-28 Tibo Sottiaux wrote that 'some users had stacked up to three banked resets already'. A reset is only consumed when it actually refreshes a window, so an unused one survives a global reset. What the official docs say - "Banked rate-limit resets are usable for 30 days after they're granted." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Eligibility, affected usage limits, delivery timing, and expiration vary by offer, plan, workspace, and region." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Check the expiration shown in your account or offer. Once an unused reset expires, it cannot be restored or reissued." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Banked resets are not cash, API credits, or transferable credit balances." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "The reset is consumed only when it successfully refreshes at least one eligible usage window." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "some users had stacked up to three banked resets already" — Tibo Sottiaux on X, 2026-06-28 (users had stacked up to three banked resets) (https://x.com/thsottiaux/status/2071381664853319742) - "Type /usage to open the usage menu. Choose whether to show token activity or redeem an available earned reset." — Codex CLI docs: Developer commands (/usage) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) How it is counted | Question | Answer | Source | | --- | --- | --- | | How long is a banked reset usable? | 30 days after it is granted (referral-promotion terms, June 2026) | [1] | | Same for every offer? | No. Expiration varies by offer, plan, workspace and region; check the date shown in your account | [2] | | What if it expires unused? | It cannot be restored or reissued | [2] | | Can I hold more than one? | Yes. Users had stacked up to three by 2026-06-28 | [3] | | Is it money or credit? | No. Not cash, API credits or a transferable balance | [2] | | Does a global reset use it up? | No. It is consumed only when it refreshes at least one eligible window | [2] | FAQ: Q: Where do I see when my banked reset expires? A: The help center says to 'Check the expiration shown in your account or offer.' In the ChatGPT desktop app and on web it is under Settings → Usage next to the reset; in the CLI, /usage opens the menu that lets you redeem an available reset. Q: Will OpenAI replace a reset I let expire? A: No. The help center says 'Once an unused reset expires, it cannot be restored or reissued.' Support does not offer replacement resets, credits or refunds because an unused banked reset expired. Q: Can I keep more than one banked reset? A: Yes. Tibo Sottiaux wrote on 2026-06-28 that 'some users had stacked up to three banked resets already' and could apply them on their own schedule. Each one has its own expiry. Q: If a global reset happens, do I lose my saved reset? A: No. The help center says 'The reset is consumed only when it successfully refreshes at least one eligible usage window.' If there is nothing to reset, it stays available, though its expiry date keeps running. Sources: 1. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 2. OpenAI Help Center: How banked Codex resets work — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 3. Tibo Sottiaux on X, 2026-06-28 (users had stacked up to three banked resets) — https://x.com/thsottiaux/status/2071381664853319742 (verified 2026-09-17) 4. Codex CLI docs: Developer commands (/usage) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) ### Codex 重置卡会过期吗？ URL: https://didreset.com/zh/guides/do-banked-resets-expire Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 会。Codex 定价页写明，重置卡在发放后 30 天内可用，这句话出现在 2026 年 6 月推荐活动的条款里；帮助中心则说有效期会因活动、计划、工作区和地区而不同，所以要看账号里显示的日期，并明说：未使用的重置一旦过期，无法恢复或重新发放。重置卡不是现金、API 积分或可转让余额；Tibo Sottiaux 提过有用户攒了三张。 定义 唯一公布的数字是 30 天。定价页说重置卡在发放后 30 天内可用。这一句写在 2026 年 6 月 11 日至 24 日推荐活动的条款里，描述的是那批发放，不一定适用于所有活动。 帮助中心给出的是按活动而定的通则：资格、影响的限额、发放时间和有效期会因活动、计划、工作区和地区而不同。它让你查看账号或活动里显示的到期时间，并直说后果：未使用的重置一旦过期，无法恢复或重新发放。 重置卡是促销权益，不是钱：帮助中心明确说它不是现金、API 积分或可转让的积分余额。你可以同时持有多张；Tibo Sottiaux 在 2026-06-28 写道有用户已经攒了三张。重置只有在真的刷新了某个窗口时才会被消耗，所以没用掉的那张在全员重置之后仍然保留。 官方文档怎么说 - "Banked rate-limit resets are usable for 30 days after they're granted." — Codex docs: Pricing (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Eligibility, affected usage limits, delivery timing, and expiration vary by offer, plan, workspace, and region." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Check the expiration shown in your account or offer. Once an unused reset expires, it cannot be restored or reissued." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Banked resets are not cash, API credits, or transferable credit balances." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "The reset is consumed only when it successfully refreshes at least one eligible usage window." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "some users had stacked up to three banked resets already" — Tibo Sottiaux on X, 2026-06-28 (users had stacked up to three banked resets) (https://x.com/thsottiaux/status/2071381664853319742) - "Type /usage to open the usage menu. Choose whether to show token activity or redeem an available earned reset." — Codex CLI docs: Developer commands (/usage) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) 怎么算 | 问题 | 答案 | 来源 | | --- | --- | --- | | 重置卡能用多久？ | 发放后 30 天内（2026 年 6 月推荐活动条款） | [1] | | 每次活动都一样吗？ | 不一样。有效期因活动、计划、工作区和地区而不同；以账号里显示的日期为准 | [2] | | 过期没用会怎样？ | 无法恢复或重新发放 | [2] | | 能同时持有多张吗？ | 能。到 2026-06-28 已有用户攒了三张 | [3] | | 它是钱或积分吗？ | 不是。不是现金、API 积分或可转让余额 | [2] | | 全员重置会把它用掉吗？ | 不会。只有真的刷新了至少一个窗口时才会被消耗 | [2] | FAQ: Q: 在哪里能看到我的重置卡什么时候过期？ A: 帮助中心让你查看账号或活动里显示的到期时间。ChatGPT 桌面应用与网页版在 Settings → Usage 里重置卡旁边显示；CLI 里输入 /usage 会打开可以兑换重置的菜单。 Q: 重置卡过期了，OpenAI 会补发吗？ A: 不会。帮助中心说未使用的重置一旦过期，无法恢复或重新发放。客服不会因为一张未用的重置卡过期而提供替代重置、积分或退款。 Q: 可以同时攒多张重置卡吗？ A: 可以。Tibo Sottiaux 在 2026-06-28 写道有用户已经攒了三张，可以按自己的节奏使用。每一张有各自的到期时间。 Q: 遇到全员重置，我存着的重置卡会没掉吗？ A: 不会。帮助中心说重置只有在成功刷新至少一个符合条件的窗口时才会被消耗。没有东西可重置时它保持可用，不过到期时间照常走。 Sources: 1. Codex docs: Pricing (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 2. OpenAI Help Center: How banked Codex resets work — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 3. Tibo Sottiaux on X, 2026-06-28 (users had stacked up to three banked resets) — https://x.com/thsottiaux/status/2071381664853319742 (verified 2026-09-17) 4. Codex CLI docs: Developer commands (/usage) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) ### Does upgrading an older account to Pro grant past Codex banked resets? URL: https://didreset.com/guides/banked-reset-existing-account-upgrade Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 OpenAI has not published this. The help center says 'Banked resets are promotional benefits. Eligibility and any qualifying cutoff depend on the specific offer or announcement' and 'Future resets are not guaranteed.' Each drop had its own cutoff: existing paid users in good standing received one on both September 3 and September 4, 2026, and Tibo Sottiaux wrote 'If you create the account or upgrade before 8pm PT you will get it too.' Nothing documents a past drop as retroactive. Definition There is no published rule that an upgrade back-issues earlier resets. The help center's position is per offer: 'Banked resets are promotional benefits. Eligibility and any qualifying cutoff depend on the specific offer or announcement.' It adds that eligibility may vary by plan, region, availability and whether your plan is new or existing, and that 'Future resets are not guaranteed.' The GPT-6 Astra launch drops show how cutoffs work. Existing paid users 'in good standing received a banked reset both on September 3rd and September 4th', and new accounts qualified if created before 10 PM PT on September 3 or 8 PM PT on September 4, 2026. Tibo Sottiaux's post of 2026-09-05 00:39 UTC (the evening of September 4 Pacific time) said 'If you create the account or upgrade before 8pm PT you will get it too.' Upgrading before that cutoff counted for that drop only. The help center also limits what a later request can do: 'The daily reset offers covered the broader Astra launch delay. Individual access issues after launch do not qualify you for an additional reset.' This site therefore treats past drops as not retroactive and labels the question 'not published' rather than guessing. What the official docs say - "Banked resets are promotional benefits. Eligibility and any qualifying cutoff depend on the specific offer or announcement." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Future resets are not guaranteed." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "in good standing received a banked reset both on September 3rd and September 4th" — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "PS: If you create the account or upgrade before 8pm PT you will get it too. Still time!" — Tibo Sottiaux on X, posted 2026-09-05 00:39 UTC (upgrade before 8pm PT cutoff) (https://x.com/thsottiaux/status/2096035437299237298) - "The daily reset offers covered the broader Astra launch delay. Individual access issues after launch do not qualify you for an additional reset." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "open Settings → Usage in the ChatGPT desktop app and select an option such as "1 reset available" or "Full reset."" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Type /usage to open the usage menu. Choose whether to show token activity or redeem an available earned reset." — Codex CLI docs: Developer commands (/usage) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) How it is counted | Question | What OpenAI has said | Source | | --- | --- | --- | | Are past banked resets back-issued when I upgrade? | Not published. Eligibility and any cutoff depend on the specific offer or announcement | [1] | | Astra launch drops, September 3–4, 2026 | Existing paid users in good standing received one on both days; new accounts needed to exist before 10 PM PT (Sep 3) or 8 PM PT (Sep 4) | [1] | | Upgrading before a cutoff | Counted for that drop: "If you create the account or upgrade before 8pm PT you will get it too" | [2] | | Individual access issues after launch | Do not qualify you for an additional reset | [3] | | Future drops | "Future resets are not guaranteed." | [1] | FAQ: Q: If I upgrade to Pro today, do I get the September 2026 resets? A: Nothing published says so. Those drops had explicit cutoffs (10 PM PT on September 3 and 8 PM PT on September 4, 2026), and the help center says eligibility depends on 'the specific offer or announcement'. An upgrade after a cutoff is not documented as qualifying for that drop. Q: Can support issue me a reset I missed? A: The help center says 'Individual access issues after launch do not qualify you for an additional reset.' Eligibility is tied to the offer's own cutoff, not to a later request. Q: How do I know whether I already have a banked reset? A: Open Settings → Usage in the ChatGPT desktop app or on web; a saved reset appears there as an option such as '1 reset available' or 'Full reset'. In the CLI, /usage opens the menu that lets you redeem an available reset. Q: Will there be more banked resets? A: OpenAI does not promise them: 'Future resets are not guaranteed.' Each past drop came with its own announcement and cutoff. The DidReset board marks BANKED when the latest official signal is a reset card drop, so you see new ones as they are announced. Sources: 1. OpenAI Help Center: How banked Codex resets work — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. Tibo Sottiaux on X, posted 2026-09-05 00:39 UTC (upgrade before 8pm PT cutoff) — https://x.com/thsottiaux/status/2096035437299237298 (verified 2026-09-17) 3. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 4. Codex CLI docs: Developer commands (/usage) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) ### 老账号升级 Pro 会补发过往的 Codex 重置卡吗？ URL: https://didreset.com/zh/guides/banked-reset-existing-account-upgrade Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 OpenAI 未公布。帮助中心说重置卡是促销权益，资格与截止时间取决于具体活动或公告，并明确未来的重置不作保证。每次发放各有截止时间：2026 年 9 月 3 日和 4 日，状态正常的现有付费用户各拿到一张；Tibo Sottiaux 写道，在太平洋时间晚上 8 点前建号或升级也能拿到那一张。没有任何文件说过往发放会追溯补发。 定义 没有任何公开规则说升级会补发早先的重置卡。帮助中心的立场是按活动而定：重置卡是促销权益，资格与截止时间取决于具体活动或公告；资格还可能因计划、地区、供应情况以及计划是新是旧而不同，而且未来的重置不作保证。 GPT-6 Astra 发布期间的发放说明了截止时间怎么起作用。状态正常的现有付费用户在 9 月 3 日和 9 月 4 日各收到一张；新账号要在 2026 年 9 月 3 日太平洋时间晚上 10 点前或 9 月 4 日晚上 8 点前建立才符合条件。Tibo Sottiaux 在 2026-09-05 00:39 UTC（太平洋时间 9 月 4 日傍晚）发帖说，在晚上 8 点前建号或升级也能拿到。截止前升级只算那一次发放。 帮助中心还限定了事后申请能做什么：每日重置活动覆盖的是 Astra 发布整体延迟；发布后个人的访问问题不构成额外重置的资格。因此本站把过往发放视为不追溯，把这个问题标为「未公布」而不去猜。 官方文档怎么说 - "Banked resets are promotional benefits. Eligibility and any qualifying cutoff depend on the specific offer or announcement." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Future resets are not guaranteed." — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "in good standing received a banked reset both on September 3rd and September 4th" — OpenAI Help Center: How banked Codex resets work (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "PS: If you create the account or upgrade before 8pm PT you will get it too. Still time!" — Tibo Sottiaux on X, posted 2026-09-05 00:39 UTC (upgrade before 8pm PT cutoff) (https://x.com/thsottiaux/status/2096035437299237298) - "The daily reset offers covered the broader Astra launch delay. Individual access issues after launch do not qualify you for an additional reset." — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "open Settings → Usage in the ChatGPT desktop app and select an option such as "1 reset available" or "Full reset."" — OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Type /usage to open the usage menu. Choose whether to show token activity or redeem an available earned reset." — Codex CLI docs: Developer commands (/usage) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) 怎么算 | 问题 | OpenAI 说过什么 | 来源 | | --- | --- | --- | | 升级后会补发过往的重置卡吗？ | 未公布。资格与截止时间取决于具体活动或公告 | [1] | | 2026 年 9 月 3–4 日的 Astra 发布发放 | 状态正常的现有付费用户两天各一张；新账号须在 9 月 3 日 22:00 PT 或 9 月 4 日 20:00 PT 前建立 | [1] | | 在截止前升级 | 算那一次发放：截止前建号或升级也能拿到 | [2] | | 发布后个人的访问问题 | 不构成额外重置的资格 | [3] | | 以后还会发吗 | 未来的重置不作保证 | [1] | FAQ: Q: 我今天升级 Pro，能拿到 2026 年 9 月那两张重置卡吗？ A: 没有公开说法支持。那两次发放有明确截止时间（2026 年 9 月 3 日 22:00 PT 与 9 月 4 日 20:00 PT），帮助中心说资格取决于具体活动或公告。截止之后升级没有被记录为符合那次发放的条件。 Q: 客服能补给我错过的重置卡吗？ A: 帮助中心说发布后个人的访问问题不构成额外重置的资格。资格绑定在活动自己的截止时间上，不取决于事后的申请。 Q: 怎么知道我账号里有没有重置卡？ A: 在 ChatGPT 桌面应用或网页版打开 Settings → Usage，有存着的重置时那里会出现「1 reset available」或「Full reset」这样的选项。CLI 里输入 /usage 会打开可以兑换的菜单。 Q: 以后还会发重置卡吗？ A: OpenAI 不作承诺：未来的重置不作保证。过往每次发放都有自己的公告和截止时间。DidReset 出发板在最新官方信号是重置卡发放时会显示 BANKED，新一轮发放公告出来你就能看到。 Sources: 1. OpenAI Help Center: How banked Codex resets work — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. Tibo Sottiaux on X, posted 2026-09-05 00:39 UTC (upgrade before 8pm PT cutoff) — https://x.com/thsottiaux/status/2096035437299237298 (verified 2026-09-17) 3. OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codex — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 4. Codex CLI docs: Developer commands (/usage) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) ### Do Claude Code's 5-hour session and weekly limits reset together? URL: https://didreset.com/guides/claude-code-5-hour-and-weekly-limits Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 No. Anthropic's help center says 'Your session-based usage limit will reset every five hours', while the weekly limit 'resets at a fixed time each week that is assigned to your account', regardless of when you start using Claude. Both apply on Pro and Max, are shared across claude.ai, Claude Code and the IDE, and Anthropic publishes no message counts. Weekly limits were announced on 2025-07-28 for late August. Run /usage in Claude Code to see both bars. Definition Two clocks. The Max plan article says 'Your session-based usage limit will reset every five hours.' Pro and Max 'also have a weekly usage limit that applies across all models', and 'The weekly limit resets at a fixed time each week that is assigned to your account.' The weekly clock does not move with your activity: 'Your reset day and time stay the same regardless of when you start using Claude or when your subscription begins'. One pool. 'Both Pro and Max plans offer usage limits that are shared across Claude and Claude Code', and IDE usage counts toward the same usage limits. Max 5x 'includes five times the Pro plan's per-session usage allowance' and Max 20x twenty times; Settings > Usage shows a separate weekly bar 'for Opus only and all other models'. Buying usage credits does not move either clock: 'Usage credits don't affect this reset timing.' No published counts. Anthropic announced the weekly limits on 2025-07-28: 'We're rolling out new weekly rate limits for Claude Pro and Max in late August.' It publishes no message numbers; the Pro article says usage varies with message length, the length of your current conversation, and the model or feature you use. In Claude Code, /usage shows the plan usage bars. What the official docs say - "Your session-based usage limit will reset every five hours." — Claude Help Center: What is the Max plan? (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "The weekly limit resets at a fixed time each week that is assigned to your account." — Claude Help Center: What is the Max plan? (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "Your reset day and time stay the same regardless of when you start using Claude or when your subscription begins" — Claude Help Center: What is the Max plan? (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "Both Pro and Max plans offer usage limits that are shared across Claude and Claude Code" — Claude Help Center: Use Claude Code with your Pro or Max plan (https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) - "Check when your plan's weekly usage limit resets for Opus only and all other models." — Claude Help Center: Usage limit best practices (https://support.claude.com/en/articles/9797557-usage-limit-best-practices) - "We're rolling out new weekly rate limits for Claude Pro and Max in late August." — Anthropic on X, 2025-07-28 (weekly rate limits announced) (https://x.com/AnthropicAI/status/1949898502688903593) - "Usage credits don't affect this reset timing." — Claude Help Center: Manage usage credits for paid Claude plans (https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans) - "Pro plans also have a weekly usage limit that applies across all models." — Claude Help Center: What is the Pro plan? (https://support.claude.com/en/articles/8325606-what-is-the-pro-plan) How it is counted | Question | Claude answer | Source | | --- | --- | --- | | How often does the session limit reset? | Every five hours | [1] | | When does the weekly limit reset? | At a fixed time each week assigned to your account; it stays the same regardless of when you start | [1] [6] | | Do claude.ai, Claude Code and the IDE share it? | Yes. Shared across Claude and Claude Code; IDE usage counts toward the same limits | [2] | | Max 5x / Max 20x | Five times / twenty times the Pro per-session allowance | [1] | | Opus | A separate weekly bar for Opus only, next to all other models | [3] | | Do usage credits move the clock? | No. "Usage credits don't affect this reset timing." | [5] | | Published message counts? | Not published. Usage varies with message length, conversation length and the model or feature | [6] | | Where to check | /usage in Claude Code shows the plan usage bars; Settings > Usage on claude.ai | [7] [3] | FAQ: Q: When did Claude weekly limits start? A: Anthropic posted on 2025-07-28: 'We're rolling out new weekly rate limits for Claude Pro and Max in late August.' The same post estimated they would apply to less than 5% of subscribers based on usage at the time. Q: Does buying usage credits reset my session window? A: No. The help center says 'Your plan's included usage limit will reset every five hours once you reach it. Usage credits don't affect this reset timing.' Credits let you keep working past the limit; they do not move the clock. Q: Are Claude Code and claude.ai separate limits? A: No. Anthropic says 'Both Pro and Max plans offer usage limits that are shared across Claude and Claude Code', so all activity in both tools counts against the same session and weekly limits. IDE usage counts toward the same limits too. Q: Where do I see my weekly reset time? A: In claude.ai, Settings > Usage shows when your weekly limit resets 'for Opus only and all other models'. In Claude Code, /usage shows the plan usage bars. That time is per account; the resets on the DidReset board are account-wide announcements by Anthropic. Sources: 1. Claude Help Center: What is the Max plan? — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 2. Claude Help Center: Use Claude Code with your Pro or Max plan — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 3. Claude Help Center: Usage limit best practices — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 4. Anthropic on X, 2025-07-28 (weekly rate limits announced) — https://x.com/AnthropicAI/status/1949898502688903593 (verified 2026-09-17) 5. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 6. Claude Help Center: What is the Pro plan? — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 7. Claude Code docs: Manage costs effectively — https://code.claude.com/docs/en/costs (verified 2026-09-17) ### Claude Code 的 5 小时窗口和周限额是一起重置的吗？ URL: https://didreset.com/zh/guides/claude-code-5-hour-and-weekly-limits Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 不是。Anthropic 帮助中心说会话限额每 5 小时重置一次，周限额则在分配给你账号的每周固定时间重置，与你何时开始用无关。两者在 Pro 与 Max 上都适用，claude.ai、Claude Code 与 IDE 共用一份额度，没有公布消息条数。在 Claude Code 里运行 /usage 可看到两条额度条。 定义 两个时钟。Max 计划文章说基于会话的用量限额每 5 小时重置一次。Pro 与 Max 还有一个适用于所有模型的周限额，周限额在分配给你账号的每周固定时间重置。周时钟不随你的活动移动：无论你何时开始使用 Claude、订阅何时开始，重置的日期和时间都保持不变。 一个池子。Pro 与 Max 的用量限额在 Claude 与 Claude Code 之间共用，IDE 里的用量也计入同一限额。Max 5x 包含 Pro 计划每会话额度的 5 倍，Max 20x 为 20 倍；Settings > Usage 里分别显示「仅 Opus」与「其他所有模型」的周额度条。购买用量积分不会移动任何一个时钟：用量积分不影响重置时间。 没有公布条数。Anthropic 在 2025-07-28 宣布周限额：8 月下旬为 Claude Pro 与 Max 推出新的周速率限制。它没有公布消息条数；Pro 计划文章说用量随消息长度、当前对话的长度以及所用模型或功能而变化。在 Claude Code 里，/usage 显示计划用量条。 官方文档怎么说 - "Your session-based usage limit will reset every five hours." — Claude Help Center: What is the Max plan? (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "The weekly limit resets at a fixed time each week that is assigned to your account." — Claude Help Center: What is the Max plan? (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "Your reset day and time stay the same regardless of when you start using Claude or when your subscription begins" — Claude Help Center: What is the Max plan? (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "Both Pro and Max plans offer usage limits that are shared across Claude and Claude Code" — Claude Help Center: Use Claude Code with your Pro or Max plan (https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) - "Check when your plan's weekly usage limit resets for Opus only and all other models." — Claude Help Center: Usage limit best practices (https://support.claude.com/en/articles/9797557-usage-limit-best-practices) - "We're rolling out new weekly rate limits for Claude Pro and Max in late August." — Anthropic on X, 2025-07-28 (weekly rate limits announced) (https://x.com/AnthropicAI/status/1949898502688903593) - "Usage credits don't affect this reset timing." — Claude Help Center: Manage usage credits for paid Claude plans (https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans) - "Pro plans also have a weekly usage limit that applies across all models." — Claude Help Center: What is the Pro plan? (https://support.claude.com/en/articles/8325606-what-is-the-pro-plan) 怎么算 | 问题 | Claude 的规则 | 来源 | | --- | --- | --- | | 会话限额多久重置一次？ | 每 5 小时 | [1] | | 周限额什么时候重置？ | 在分配给你账号的每周固定时间；无论你何时开始用都不变 | [1] [6] | | claude.ai、Claude Code、IDE 共用吗？ | 共用。Claude 与 Claude Code 共用限额；IDE 用量计入同一限额 | [2] | | Max 5x / Max 20x | Pro 每会话额度的 5 倍 / 20 倍 | [1] | | Opus | 单独一条「仅 Opus」的周额度条，与其他所有模型并列 | [3] | | 用量积分会移动时钟吗？ | 不会。用量积分不影响重置时间 | [5] | | 公布过消息条数吗？ | 未公布。用量随消息长度、对话长度、模型或功能而变化 | [6] | | 在哪里看 | Claude Code 里 /usage 显示计划用量条；claude.ai 在 Settings > Usage | [7] [3] | FAQ: Q: Claude 的周限额是什么时候开始的？ A: Anthropic 在 2025-07-28 发帖：8 月下旬为 Claude Pro 与 Max 推出新的周速率限制。同一条帖子估计按当时用量，它会影响不到 5% 的订阅者。 Q: 购买用量积分会重置我的会话窗口吗？ A: 不会。帮助中心说计划内的用量限额在用完后每 5 小时重置一次，用量积分不影响这个重置时间。积分让你在触限后继续工作，但不会移动时钟。 Q: Claude Code 和 claude.ai 是分开的限额吗？ A: 不是。Anthropic 说 Pro 与 Max 的用量限额在 Claude 与 Claude Code 之间共用，两个工具里的所有活动都计入同一个会话限额和周限额。IDE 用量也计入同一限额。 Q: 在哪里看我的周重置时间？ A: 在 claude.ai 里，Settings > Usage 会显示「仅 Opus」与「其他所有模型」的周限额何时重置。在 Claude Code 里，/usage 显示计划用量条。这个时间按账号各自计算；DidReset 出发板上的重置是 Anthropic 的全员公告。 Sources: 1. Claude Help Center: What is the Max plan? — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 2. Claude Help Center: Use Claude Code with your Pro or Max plan — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 3. Claude Help Center: Usage limit best practices — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 4. Anthropic on X, 2025-07-28 (weekly rate limits announced) — https://x.com/AnthropicAI/status/1949898502688903593 (verified 2026-09-17) 5. Claude Help Center: Manage usage credits for paid Claude plans — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 6. Claude Help Center: What is the Pro plan? — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 7. Claude Code docs: Manage costs effectively — https://code.claude.com/docs/en/costs (verified 2026-09-17) ### Does Grok Bot usage count toward the SuperGrok weekly limit? URL: https://didreset.com/guides/grok-bot-usage-limits Language: en · Kind: Rule · Published: 2026-09-18 · Updated: 2026-09-18 No. xAI's 2026-08-26 announcement says 'Grok Bot comes with its own usage, separate from your Grok and Cursor plans', and the Grok Bot docs state 'Usage resets weekly.' It is included with paid individual Cursor plans, or by linking a SuperGrok, SuperGrok Plus or SuperGrok Heavy subscription; with both, 'Grok Bot uses whichever has more usage'. xAI publishes no per-plan numbers. The Grok chat pool is separate and resets weekly on the schedule shown in Settings → Usage. Definition Grok Bot has its own allowance. The xAI news post of 2026-08-26 says 'Grok Bot comes with its own usage, separate from your Grok and Cursor plans, so anything you hand off to a Bot won't count against your existing usage.' The docs add that 'Grok Bot subscriptions include weekly usage; eligible accounts can add on-demand usage billed from model and token cost', and simply that 'Usage resets weekly.' Access comes through Cursor or SuperGrok. The overview says Grok Bot is included with every paid individual Cursor plan and with the Cursor Teams plan, or by linking an individual SuperGrok, SuperGrok Plus or SuperGrok Heavy subscription. 'If you have both a Cursor and a SuperGrok subscription, Grok Bot uses whichever has more usage.' On 2026-08-26 the @bot account posted 'All SuperGrok and Cursor Pro subscribers now have access to Grok Bot. We're also resetting weekly usage limits for all users.' That kind of account-wide reset is what this board records for Grok. xAI publishes no numbers, and this site does not estimate them. When usage is exhausted or an on-demand spending limit is reached, the docs say to 'review usage and billing from the account access page or Usage & Billing'. The Grok chat pool is a different thing: it 'resets every week on a schedule shown in the Usage tab in Settings', paid features pause 'until your weekly limit resets', and Extra Usage Credits 'are used after your plan's included weekly usage is exhausted'. What the official docs say - "Grok Bot comes with its own usage, separate from your Grok and Cursor plans" — xAI news, 2026-08-26: Grok Bot is now included with more plans (https://x.ai/news/grok-bot-more-plans) - "Usage resets weekly." — xAI docs: Grok Bot overview (https://docs.x.ai/grok-bot/overview) - "Grok Bot subscriptions include weekly usage; eligible accounts can add on-demand usage billed from model and token cost." — xAI docs: Grok Bot frequently asked questions (https://docs.x.ai/grok-bot/faq) - "If you have both a Cursor and a SuperGrok subscription, Grok Bot uses whichever has more usage." — xAI docs: Grok Bot frequently asked questions (https://docs.x.ai/grok-bot/faq) - "All SuperGrok and Cursor Pro subscribers now have access to Grok Bot. We're also resetting weekly usage limits for all users." — Grok Bot (@bot) on X, 2026-08-26 (access expanded, weekly limits reset for all users) (https://x.com/bot/status/2092664183334846840) - "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." — xAI docs: FAQ – Grok website and apps (https://docs.x.ai/grok/faq) - "When your weekly usage limit is met, paid features will pause until your weekly limit resets." — xAI docs: FAQ – Grok website and apps (https://docs.x.ai/grok/faq) How it is counted | Question | Grok Bot answer | Source | | --- | --- | --- | | Does Grok Bot draw from the SuperGrok or Cursor pool? | No. Its own usage, separate from your Grok and Cursor plans | [3] | | How often does it reset? | Weekly | [1] | | Published number? | Not published. Subscriptions include weekly usage; eligible accounts can add on-demand usage billed from model and token cost | [2] | | Which plans include it? | Paid individual Cursor plans and Cursor Teams; or a linked SuperGrok, SuperGrok Plus or SuperGrok Heavy subscription | [1] [3] [4] | | Both Cursor and SuperGrok? | Grok Bot uses whichever has more usage | [2] | | Usage exhausted? | Review usage and billing from the account access page or Usage & Billing | [5] | | Grok chat (SuperGrok) pool | Separate. Resets weekly on the schedule in Settings → Usage; paid features pause when met; top-up credits used after included usage and expire one year after purchase | [6] | FAQ: Q: When does Grok Bot usage reset? A: The Grok Bot docs say only 'Usage resets weekly.' xAI has not published a day or hour. For your account's usage and billing, the troubleshooting page points to the account access page or Usage & Billing. Q: Which plans include Grok Bot? A: Per the docs overview, every paid individual Cursor plan and the Cursor Teams plan, or a linked individual SuperGrok, SuperGrok Plus or SuperGrok Heavy subscription. xAI's 2026-08-26 post confirmed 'All SuperGrok and Cursor Pro subscribers now have access to Grok Bot.' Q: Does using Grok Bot use up my Grok chat or Cursor allowance? A: No. xAI says 'Grok Bot comes with its own usage, separate from your Grok and Cursor plans, so anything you hand off to a Bot won't count against your existing usage.' If you have both Cursor and SuperGrok, Grok Bot uses whichever has more usage. Q: Has xAI ever reset everyone at once? A: Yes. On 2026-08-26 the @bot account posted 'We're also resetting weekly usage limits for all users.' Account-wide resets like that are what the DidReset board records for Grok; your own weekly window otherwise refreshes on its own schedule. Sources: 1. xAI docs: Grok Bot overview — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 2. xAI docs: Grok Bot frequently asked questions — https://docs.x.ai/grok-bot/faq (verified 2026-09-17) 3. xAI news, 2026-08-26: Grok Bot is now included with more plans — https://x.ai/news/grok-bot-more-plans (verified 2026-09-17) 4. Grok Bot (@bot) on X, 2026-08-26 (access expanded, weekly limits reset for all users) — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 5. xAI docs: Grok Bot troubleshooting — https://docs.x.ai/grok-bot/troubleshooting (verified 2026-09-17) 6. xAI docs: FAQ – Grok website and apps — https://docs.x.ai/grok/faq (verified 2026-09-17) ### Grok Bot 的用量算不算 SuperGrok 的周限额？ URL: https://didreset.com/zh/guides/grok-bot-usage-limits Language: zh · Kind: 规则 · Published: 2026-09-18 · Updated: 2026-09-18 不算。xAI 在 2026-08-26 的公告里说 Grok Bot 自带额度，与你的 Grok 和 Cursor 计划分开；Grok Bot 文档写明用量每周重置。它包含在付费 Cursor 计划里，或绑定 SuperGrok 系列订阅获得；两者都有时用余额更多的那个。xAI 没有公布数字。Grok 聊天的额度池另算，每周按设置里的时间表重置。 定义 Grok Bot 有自己的额度。xAI 2026-08-26 的新闻页说 Grok Bot 自带额度，与你的 Grok 和 Cursor 计划分开，交给 Bot 的任何工作都不会占用你现有的额度。文档补充说 Grok Bot 订阅包含每周额度，符合条件的账号可以按模型与 token 成本追加按需用量，并只说用量每周重置。 访问权来自 Cursor 或 SuperGrok。概览页说 Grok Bot 包含在每个付费的 Cursor 个人计划和 Cursor Teams 计划里，或通过绑定个人的 SuperGrok、SuperGrok Plus 或 SuperGrok Heavy 订阅获得。如果你同时有 Cursor 和 SuperGrok 订阅，Grok Bot 用剩余额度更多的那一个。2026-08-26，@bot 账号发帖：所有 SuperGrok 与 Cursor Pro 订阅者现在都能用 Grok Bot，同时为所有用户重置周限额。这种全员重置就是本站出发板为 Grok 记录的内容。 xAI 没有公布数字，本站也不估算。用量耗尽或达到按需消费上限时，文档说到账号访问页或 Usage & Billing 查看用量与账单。Grok 聊天的额度池是另一回事：它按设置里 Usage 标签页显示的时间表每周重置，触限后付费功能暂停直到周限额重置，额外用量积分在计划包含的周用量用完后才使用。 官方文档怎么说 - "Grok Bot comes with its own usage, separate from your Grok and Cursor plans" — xAI news, 2026-08-26: Grok Bot is now included with more plans (https://x.ai/news/grok-bot-more-plans) - "Usage resets weekly." — xAI docs: Grok Bot overview (https://docs.x.ai/grok-bot/overview) - "Grok Bot subscriptions include weekly usage; eligible accounts can add on-demand usage billed from model and token cost." — xAI docs: Grok Bot frequently asked questions (https://docs.x.ai/grok-bot/faq) - "If you have both a Cursor and a SuperGrok subscription, Grok Bot uses whichever has more usage." — xAI docs: Grok Bot frequently asked questions (https://docs.x.ai/grok-bot/faq) - "All SuperGrok and Cursor Pro subscribers now have access to Grok Bot. We're also resetting weekly usage limits for all users." — Grok Bot (@bot) on X, 2026-08-26 (access expanded, weekly limits reset for all users) (https://x.com/bot/status/2092664183334846840) - "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." — xAI docs: FAQ – Grok website and apps (https://docs.x.ai/grok/faq) - "When your weekly usage limit is met, paid features will pause until your weekly limit resets." — xAI docs: FAQ – Grok website and apps (https://docs.x.ai/grok/faq) 怎么算 | 问题 | Grok Bot 的规则 | 来源 | | --- | --- | --- | | Grok Bot 会占 SuperGrok 或 Cursor 的额度池吗？ | 不会。自带额度，与你的 Grok 和 Cursor 计划分开 | [3] | | 多久重置一次？ | 每周 | [1] | | 公布过数字吗？ | 未公布。订阅包含每周额度；符合条件的账号可按模型与 token 成本追加按需用量 | [2] | | 哪些计划包含它？ | 付费的 Cursor 个人计划与 Cursor Teams；或绑定 SuperGrok、SuperGrok Plus、SuperGrok Heavy 订阅 | [1] [3] [4] | | 同时有 Cursor 和 SuperGrok？ | Grok Bot 用剩余额度更多的那一个 | [2] | | 额度用完了？ | 到账号访问页或 Usage & Billing 查看用量与账单 | [5] | | Grok 聊天（SuperGrok）的额度池 | 分开的。按 Settings → Usage 里的时间表每周重置；触限后付费功能暂停；额外积分在包含用量用完后使用，购买后一年过期 | [6] | FAQ: Q: Grok Bot 的用量什么时候重置？ A: Grok Bot 文档只说用量每周重置。xAI 没有公布具体的星期几或几点。要看自己账号的用量与账单，故障排查页指向账号访问页或 Usage & Billing。 Q: 哪些计划包含 Grok Bot？ A: 按文档概览页，每个付费的 Cursor 个人计划和 Cursor Teams 计划，或绑定的个人 SuperGrok、SuperGrok Plus、SuperGrok Heavy 订阅。xAI 在 2026-08-26 的帖子里确认所有 SuperGrok 与 Cursor Pro 订阅者现在都能用 Grok Bot。 Q: 用 Grok Bot 会消耗我 Grok 聊天或 Cursor 的额度吗？ A: 不会。xAI 说 Grok Bot 自带额度，与你的 Grok 和 Cursor 计划分开，交给 Bot 的任何工作都不会占用你现有的额度。同时有 Cursor 和 SuperGrok 时，Grok Bot 用剩余额度更多的那一个。 Q: xAI 给所有人一起重置过吗？ A: 有。2026-08-26，@bot 账号发帖说同时为所有用户重置周限额。这类全员重置就是 DidReset 出发板为 Grok 记录的内容；除此之外，你自己的周窗口按各自的时间表刷新。 Sources: 1. xAI docs: Grok Bot overview — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 2. xAI docs: Grok Bot frequently asked questions — https://docs.x.ai/grok-bot/faq (verified 2026-09-17) 3. xAI news, 2026-08-26: Grok Bot is now included with more plans — https://x.ai/news/grok-bot-more-plans (verified 2026-09-17) 4. Grok Bot (@bot) on X, 2026-08-26 (access expanded, weekly limits reset for all users) — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 5. xAI docs: Grok Bot troubleshooting — https://docs.x.ai/grok-bot/troubleshooting (verified 2026-09-17) 6. xAI docs: FAQ – Grok website and apps — https://docs.x.ai/grok/faq (verified 2026-09-17) ### How to check your remaining Codex usage with /status and /usage URL: https://didreset.com/guides/check-codex-usage-status Language: en · Kind: How-to · Published: 2026-09-18 · Updated: 2026-09-18 Type /status inside an active Codex CLI session. OpenAI's pricing page says "If you want to see your remaining limits during an active Codex CLI session, you can use /status." The output shows a "5h limit" bar and a "Weekly limit" bar, each with a percentage left and a reset time. /usage opens the usage menu for token activity or an earned reset. On web and in the desktop app, open Settings → Usage. Definition Codex has two slash commands that touch usage. The official command reference describes /status as "Display session configuration and token usage" and /usage as "View account token usage or use a rate-limit reset." Both are typed at the Codex prompt inside the interactive session, not as arguments on your shell command line. The /status box prints one bar per window. A user-posted example in the official Codex repository (issue #29948) reads "5h limit: [████████████████░░░░] 82% left (resets 22:41)" and "Weekly limit: [███████████████████░] 97% left (resets 17:41 on 1 Jul)". The time after "resets" is your own window clock, not an account-wide reset. Outside the CLI, OpenAI's help center says to "Check Settings → Usage before a large task to see your remaining allowance, usage windows and reset times." The published per-plan message estimates "are not fixed message limits; check your usage dashboard for current limits and reset times." Weekly numbers are not published at all. What the official docs say - "If you want to see your remaining limits during an active Codex CLI session, you can use /status." — Codex pricing and usage limits (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "/status — Display session configuration and token usage" — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) - "/usage — View account token usage or use a rate-limit reset." — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) - "In an active Codex CLI session, enter /status." — Using Codex with your ChatGPT plan (OpenAI Help Center) (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) - "Check Settings → Usage before a large task to see your remaining allowance, usage windows and reset times." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) Steps 1. Open an active Codex session: Start Codex in your project and wait for the prompt. Slash commands such as /status work inside the interactive session (the TUI); the official command reference lists them under the CLI's developer commands. 2. Type /status: Enter /status at the Codex prompt. The box shows session configuration and token usage, including the rate-limit bars for your account. OpenAI's help center: "In an active Codex CLI session, enter /status." 3. Read the "5h limit" line: The first bar shows the percentage left in your 5-hour window and the time it refreshes, for example "82% left (resets 22:41)". That time is your own window; it is not a Codex-wide reset. 4. Read the "Weekly limit" line: The second bar shows what remains of the weekly window and when it refreshes, for example "97% left (resets 17:41 on 1 Jul)". OpenAI publishes no weekly number, so this bar is the only figure you get. 5. Type /usage for token activity: /usage opens the usage menu: "Choose whether to show token activity or redeem an available earned reset." Token activity can be inspected daily, weekly or cumulatively from inside the TUI. 6. Check outside the CLI: In the ChatGPT desktop app or on web, open Settings → Usage to see remaining allowance, usage windows and reset times. Local messages and cloud chats share your plan's usage allowance, so the numbers are the same pool. 7. Put the reset time on a countdown: Paste the time after "resets" into the window tracker on this site for a countdown and a calendar file. An account-wide reset is a separate event: it is announced by OpenAI and shown on the board, not in /status. FAQ: Q: Do I run "codex /status" from the shell? A: No. /status is a slash command typed at the Codex prompt inside a running session. The official command reference lists it among the interactive CLI's developer commands, together with /usage and /logout. Q: Why does the 5-hour bar show a reset time while I still have weekly usage? A: The two windows run on separate clocks. OpenAI's help center: "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." Where both apply, you need allowance remaining in both to continue. Q: Is the "resets" time in /status the same as a reset on this board? A: No. The time in /status is your personal window. The board tracks account-wide resets announced by OpenAI, such as the global reset documented for 2026-09-07, which "won't appear as a saved reset in Settings" and applies to everyone at once. Q: The dashboard did not change after a reset. Is something wrong? A: OpenAI's help center: "Your usage dashboard may take a short time to update. If the percentage does not change immediately, refresh or reopen Settings → Usage". If it still shows no change, check which window the reset was supposed to refresh. Q: Does /status still show a 5h limit on Pro? A: Tibo Sottiaux said on 2026-08-25 that OpenAI is "keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months, while the 5h limit returned for Plus. How the /status box renders a disabled window is not documented; read your own output. Sources: 1. Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) 2. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 3. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 4. Using Codex with your ChatGPT plan (OpenAI Help Center) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 5. openai/codex issue #29948: /status output with 5h limit and Weekly limit bars — https://github.com/openai/codex/issues/29948 (verified 2026-09-17) 6. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit returns for Plus "tomorrow"; not enabled for Pro — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### 怎么用 /status 和 /usage 查 Codex 剩余额度 URL: https://didreset.com/zh/guides/check-codex-usage-status Language: zh · Kind: 操作 · Published: 2026-09-18 · Updated: 2026-09-18 在正在运行的 Codex CLI 会话里输入 /status，OpenAI 定价页写明这是查剩余额度的方式。输出里有「5h limit」和「Weekly limit」两条进度条，各带剩余百分比和刷新时间。/usage 打开用量菜单，看 token 活动或兑换已获得的重置。网页版与桌面 App 在 Settings → Usage 查看。 定义 Codex 有两条和额度有关的斜杠命令。官方命令参考对 /status 的描述是 "Display session configuration and token usage"，对 /usage 的描述是 "View account token usage or use a rate-limit reset."。两条都在交互会话的 Codex 提示符下输入，不是 shell 命令行参数。 /status 的方框每个窗口一条进度条。官方 Codex 仓库 issue #29948 里用户贴出的样例是 "5h limit: [████████████████░░░░] 82% left (resets 22:41)" 和 "Weekly limit: [███████████████████░] 97% left (resets 17:41 on 1 Jul)"。"resets" 后面的时间是你自己的窗口时钟，不是全员重置。 CLI 之外，OpenAI 帮助中心建议 "Check Settings → Usage before a large task to see your remaining allowance, usage windows and reset times."。定价页公布的各计划消息数只是估计，"are not fixed message limits; check your usage dashboard for current limits and reset times."。周限额的具体数字则完全未公布。 官方文档怎么说 - "If you want to see your remaining limits during an active Codex CLI session, you can use /status." — Codex pricing and usage limits (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "/status — Display session configuration and token usage" — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) - "/usage — View account token usage or use a rate-limit reset." — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) - "In an active Codex CLI session, enter /status." — Using Codex with your ChatGPT plan (OpenAI Help Center) (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) - "Check Settings → Usage before a large task to see your remaining allowance, usage windows and reset times." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) 步骤 1. 打开一个正在运行的 Codex 会话: 在项目目录启动 Codex，等提示符出现。/status 这类斜杠命令只在交互会话（TUI）里生效；官方命令参考把它们列在 CLI 的开发者命令下。 2. 输入 /status: 在 Codex 提示符下输入 /status 回车。方框会显示会话配置和 token 用量，包括你账号的限额进度条。OpenAI 帮助中心原话："In an active Codex CLI session, enter /status." 3. 读「5h limit」这一行: 第一条进度条是 5 小时窗口的剩余百分比和刷新时间，例如 "82% left (resets 22:41)"。这个时间是你自己的窗口，不是 Codex 全员重置。 4. 读「Weekly limit」这一行: 第二条进度条是周窗口的剩余量和刷新时间，例如 "97% left (resets 17:41 on 1 Jul)"。OpenAI 没有公布周限额数字，这条进度条是你能拿到的唯一数据。 5. 输入 /usage 看 token 活动: /usage 打开用量菜单："Choose whether to show token activity or redeem an available earned reset."。token 活动可以在 TUI 里按日、按周或累计查看。 6. CLI 之外的查法: 在 ChatGPT 桌面 App 或网页版打开 Settings → Usage，查看剩余额度、用量窗口和刷新时间。本地消息和云端对话共用同一个计划额度，所以看到的是同一个池子。 7. 把刷新时间做成倒计时: 把 "resets" 后面的时间贴进本站的窗口倒计时工具，得到倒计时和日历文件。全员重置是另一件事：由 OpenAI 公告、显示在出发板上，不会出现在 /status 里。 FAQ: Q: 是在 shell 里跑 "codex /status" 吗？ A: 不是。/status 是在运行中的 Codex 会话提示符下输入的斜杠命令。官方命令参考把它和 /usage、/logout 一起列在交互式 CLI 的开发者命令里。 Q: 为什么周额度还有余量，5 小时条却显示刷新时间？ A: 两个窗口各走各的时钟。OpenAI 帮助中心："You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining."。两个窗口都适用时，两边都要有余量才能继续。 Q: /status 里的 "resets" 时间和本站出发板上的重置是一回事吗？ A: 不是。/status 里的时间是你个人的窗口。出发板追踪的是 OpenAI 公告的全员重置，例如帮助中心记载的 2026-09-07 全局重置，它 "won't appear as a saved reset in Settings"，对所有人同时生效。 Q: 重置之后仪表盘没变化，是不是出问题了？ A: OpenAI 帮助中心："Your usage dashboard may take a short time to update. If the percentage does not change immediately, refresh or reopen Settings → Usage"。如果仍然没变，先确认这次重置本来要刷新的是哪个窗口。 Q: Pro 计划的 /status 还会显示 5h limit 吗？ A: Tibo Sottiaux 在 2026-08-25 说 OpenAI 在接下来几个月 "keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions"，而 Plus 的 5 小时限制恢复。被停用的窗口在 /status 里怎么显示没有文档说明，以你自己的输出为准。 Sources: 1. Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) 2. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 3. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 4. Using Codex with your ChatGPT plan (OpenAI Help Center) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 5. openai/codex issue #29948: /status output with 5h limit and Weekly limit bars — https://github.com/openai/codex/issues/29948 (verified 2026-09-17) 6. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit returns for Plus "tomorrow"; not enabled for Pro — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### How to check your remaining Claude Code usage with /usage URL: https://didreset.com/guides/check-claude-code-usage Language: en · Kind: How-to · Published: 2026-09-18 · Updated: 2026-09-18 Run /usage inside Claude Code. The official docs say Pro and Max subscribers "see plan usage bars, activity stats, and a usage breakdown on the same screen." The same limits appear at Settings > Usage on claude.ai, which shows "How much of your plan's five-hour session limit you've used thus far, plus the amount of time remaining in the session." Anthropic's help center also names /status for monitoring your remaining allocation. Definition Claude Code meters Pro and Max usage against the same pool as claude.ai: "all activity in both tools counts against the same usage limits." /usage is the in-terminal view of that pool, and the session bar it shows is the one that "will reset every five hours." On a Pro, Max, Team, or Enterprise plan, /usage also shows a breakdown of what counts against your plan limits, with recent usage attributed to skills, subagents, plugins and individual MCP servers. That tells you what drained the window, not only how much is left. The web view at Settings > Usage lists the five-hour session used so far with time remaining, and lets you "Check when your plan's weekly usage limit resets for Opus only and all other models." The weekly limit "resets at a fixed time each week that is assigned to your account." Anthropic does not publish message counts. What the official docs say - "Subscribers see plan usage bars, activity stats, and a usage breakdown on the same screen." — Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) (https://code.claude.com/docs/en/costs) - "How much of your plan's five-hour session limit you've used thus far, plus the amount of time remaining in the session." — Usage limit best practices: Settings > Usage (Claude Help Center) (https://support.claude.com/en/articles/9797557-usage-limit-best-practices) - "Monitor your remaining allocation using the /status command" — Use Claude Code with your Pro or Max plan (Claude Help Center) (https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) - "Your session-based usage limit will reset every five hours." — What is the Max plan? (Claude Help Center) (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "Check when your plan's weekly usage limit resets for Opus only and all other models." — Usage limit best practices: Settings > Usage (Claude Help Center) (https://support.claude.com/en/articles/9797557-usage-limit-best-practices) Steps 1. Open Claude Code: Start Claude Code in your project. Slash commands are typed at the Claude Code prompt inside the session. 2. Type /usage: On a subscription plan the screen shows plan usage bars, activity stats and a usage breakdown together. The docs: "Subscribers see plan usage bars, activity stats, and a usage breakdown on the same screen." 3. Read the session and weekly bars: The session bar is the five-hour limit; the weekly bar is the cap across all models. Each reset time is your own account clock: the session "will reset every five hours", the weekly limit "resets at a fixed time each week that is assigned to your account." 4. Read the attribution: The breakdown shows "what counts against your plan limits", attributed to skills, subagents, plugins and individual MCP servers. It also flags long context or cache misses when they account for 10% or more of recent usage. 5. Cross-check at Settings > Usage on claude.ai: The web page shows how much of the five-hour session you have used and the time remaining, plus when the weekly limit resets for Opus only and for all other models. Web, desktop and Claude Code all draw from the same pool. 6. Or use /status: Anthropic's help center for Pro and Max says to "Monitor your remaining allocation using the /status command". Both /usage (Claude Code docs) and /status (help center) are official; use whichever your version shows. 7. Put the reset time on a countdown: Paste the session or weekly reset time into the window tracker on this site for a countdown and a calendar file. An account-wide reset announced by Anthropic is a separate event and shows on the board. FAQ: Q: Does chatting on claude.ai count against Claude Code? A: Yes. "Both Pro and Max plans offer usage limits that are shared across Claude and Claude Code, meaning all activity in both tools counts against the same usage limits." The help center adds that claude.ai, Claude Code and Claude Desktop all count "towards the same usage limit." Q: Why is there a separate Opus bar? A: Settings > Usage shows when the weekly limit resets "for Opus only and all other models." Claude Code's error reference has a matching message, "You've hit your Opus limit", which applies only to that model family; switching to a model outside the family with /model keeps you working. Q: Do usage credits move my reset time? A: No. "Your plan's included usage limit will reset every five hours once you reach it. Usage credits don't affect this reset timing." Q: How many messages do I get per session? A: Anthropic does not publish a count. The Pro plan page says the number "will vary based on message length ... the length of your current conversation, and the model or feature you use." Max 5x and 20x are described as five and twenty times the Pro per-session allowance, with no absolute number. Sources: 1. Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) — https://code.claude.com/docs/en/costs (verified 2026-09-17) 2. Usage limit best practices: Settings > Usage (Claude Help Center) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 3. Use Claude Code with your Pro or Max plan (Claude Help Center) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 4. What is the Max plan? (Claude Help Center) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 5. How do usage and length limits work? (Claude Help Center) — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) 6. Manage usage credits for paid Claude plans (Claude Help Center) — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 7. What is the Pro plan? (Claude Help Center) — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 8. Claude Code docs: Error reference (session, weekly, Opus and Sonnet limits) — https://code.claude.com/docs/en/errors (verified 2026-09-17) ### 怎么用 /usage 查 Claude Code 剩余额度 URL: https://didreset.com/zh/guides/check-claude-code-usage Language: zh · Kind: 操作 · Published: 2026-09-18 · Updated: 2026-09-18 在 Claude Code 里输入 /usage。官方文档写明，Pro 与 Max 订阅者在同一屏幕上看到套餐用量条、活动统计和用量归因。同样的限额也能在 claude.ai 的 Settings > Usage 看到：5 小时会话已用多少、还剩多久，以及周限额何时刷新。Anthropic 帮助中心另提到可用 /status 监控剩余额度。 定义 Claude Code 的 Pro 与 Max 用量和 claude.ai 计入同一个池子："all activity in both tools counts against the same usage limits."。/usage 就是这个池子在终端里的视图，它显示的会话条就是那个 "will reset every five hours" 的 5 小时限额。 在 Pro、Max、Team 或 Enterprise 计划上，/usage 还会显示计入套餐限额的用量归因：最近的用量按 skill、subagent、插件和各个 MCP 服务器拆开。这告诉你是什么把窗口耗掉的，而不只是还剩多少。 网页端 Settings > Usage 列出 5 小时会话已用量和剩余时间，并能 "Check when your plan's weekly usage limit resets for Opus only and all other models."。周限额 "resets at a fixed time each week that is assigned to your account."。Anthropic 不公布消息条数。 官方文档怎么说 - "Subscribers see plan usage bars, activity stats, and a usage breakdown on the same screen." — Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) (https://code.claude.com/docs/en/costs) - "How much of your plan's five-hour session limit you've used thus far, plus the amount of time remaining in the session." — Usage limit best practices: Settings > Usage (Claude Help Center) (https://support.claude.com/en/articles/9797557-usage-limit-best-practices) - "Monitor your remaining allocation using the /status command" — Use Claude Code with your Pro or Max plan (Claude Help Center) (https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) - "Your session-based usage limit will reset every five hours." — What is the Max plan? (Claude Help Center) (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "Check when your plan's weekly usage limit resets for Opus only and all other models." — Usage limit best practices: Settings > Usage (Claude Help Center) (https://support.claude.com/en/articles/9797557-usage-limit-best-practices) 步骤 1. 打开 Claude Code: 在项目目录启动 Claude Code。斜杠命令在会话内的 Claude Code 提示符下输入。 2. 输入 /usage: 订阅计划上，这一屏同时显示套餐用量条、活动统计和用量归因。文档原话："Subscribers see plan usage bars, activity stats, and a usage breakdown on the same screen." 3. 读会话条和周限额条: 会话条是 5 小时限额，周限额条是跨所有模型的上限。刷新时间都是你自己账号的时钟：会话 "will reset every five hours"，周限额 "resets at a fixed time each week that is assigned to your account." 4. 读用量归因: 归因部分显示 "what counts against your plan limits"，按 skill、subagent、插件和各个 MCP 服务器拆分。当长上下文或缓存未命中占最近用量 10% 以上时也会标出来。 5. 到 claude.ai 的 Settings > Usage 交叉核对: 网页显示 5 小时会话已用多少、还剩多长时间，以及周限额分别为 Opus 和其他模型什么时候刷新。网页、桌面和 Claude Code 用的是同一个池子。 6. 或者用 /status: Anthropic 面向 Pro 与 Max 的帮助中心文章写的是 "Monitor your remaining allocation using the /status command"。/usage（Claude Code 文档）和 /status（帮助中心）都是官方说法，你的版本显示哪个就用哪个。 7. 把刷新时间做成倒计时: 把会话或周限额的刷新时间贴进本站的窗口倒计时工具，得到倒计时和日历文件。Anthropic 公告的全员重置是另一件事，显示在出发板上。 FAQ: Q: 在 claude.ai 聊天会占用 Claude Code 的额度吗？ A: 会。"Both Pro and Max plans offer usage limits that are shared across Claude and Claude Code, meaning all activity in both tools counts against the same usage limits."。帮助中心还补充，claude.ai、Claude Code 和 Claude Desktop 的用量都计入同一个限额。 Q: 为什么会有单独的 Opus 条？ A: Settings > Usage 显示周限额 "for Opus only and all other models" 各自的刷新时间。Claude Code 的错误参考里有对应的报错 "You've hit your Opus limit"，只作用于该模型家族；用 /model 切到家族之外的模型可以继续工作。 Q: 用量积分（usage credits）会改变刷新时间吗？ A: 不会。"Your plan's included usage limit will reset every five hours once you reach it. Usage credits don't affect this reset timing." Q: 每个会话能发多少条消息？ A: Anthropic 不公布条数。Pro 计划页面说条数 "will vary based on message length ... the length of your current conversation, and the model or feature you use."。Max 5x 与 20x 的描述是 Pro 每会话额度的 5 倍和 20 倍，没有绝对数字。 Sources: 1. Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) — https://code.claude.com/docs/en/costs (verified 2026-09-17) 2. Usage limit best practices: Settings > Usage (Claude Help Center) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 3. Use Claude Code with your Pro or Max plan (Claude Help Center) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 4. What is the Max plan? (Claude Help Center) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 5. How do usage and length limits work? (Claude Help Center) — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) 6. Manage usage credits for paid Claude plans (Claude Help Center) — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 7. What is the Pro plan? (Claude Help Center) — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 8. Claude Code docs: Error reference (session, weekly, Opus and Sonnet limits) — https://code.claude.com/docs/en/errors (verified 2026-09-17) ### How to check your remaining Grok usage in Settings → Usage URL: https://didreset.com/guides/check-grok-usage Language: en · Kind: How-to · Published: 2026-09-18 · Updated: 2026-09-18 Open Settings → Usage in the Grok website or app. xAI's FAQ says the tab shows "A progress bar showing your current usage percentage" and "A percentage breakdown by product", and that "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." For Grok Bot, xAI's troubleshooting page points to Usage & Billing on the account access page. xAI publishes no per-plan numbers, and this site does not estimate them. Definition Paid Grok plans meter usage from one weekly pool shared across products; per xAI, "Different products cost different amounts depending on how much compute that product requires." The Usage tab in Settings is where the pool, its per-product split and its reset schedule live. xAI's FAQ describes this weekly pool only; it mentions no five-hour window. When the pool is used up: "When your weekly usage limit is met, paid features will pause until your weekly limit resets. You will still have access to Grok's free tier limits." Extra Usage Credits "are used after your plan's included weekly usage is exhausted. Top up credits expire one year after purchase." Grok Bot is metered separately. xAI's news post says "Grok Bot comes with its own usage, separate from your Grok and Cursor plans", and the overview states "Usage resets weekly." If usage is exhausted or an on-demand spending limit is reached, the troubleshooting page says to "review usage and billing from the account access page or Usage & Billing when that section is available." What the official docs say - "A progress bar showing your current usage percentage" — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) - "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) - "When your weekly usage limit is met, paid features will pause until your weekly limit resets." — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) - "Usage resets weekly." — xAI docs: Grok Bot overview ("Usage resets weekly.") (https://docs.x.ai/grok-bot/overview) - "review usage and billing from the account access page or Usage & Billing when that section is available" — xAI docs: Grok Bot troubleshooting (Usage & Billing) (https://docs.x.ai/grok-bot/troubleshooting) Steps 1. Open Settings in Grok: Sign in to Grok on the website or in the app and open Settings. The usage view is a tab inside Settings, not a chat command. 2. Open the Usage tab: The FAQ describes the Usage tab in Settings as the place where your weekly pool and its reset schedule are shown. 3. Read the progress bar: The bar is "A progress bar showing your current usage percentage" of the weekly pool. xAI does not publish what 100% corresponds to for any plan. 4. Read the breakdown by product: Below the bar is "A percentage breakdown by product", so you can see whether Chat, Imagine, Voice or Build drained the pool. Different products cost different amounts depending on the compute they need. 5. Note the weekly reset schedule: "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." That schedule is your own account clock; it is not the account-wide reset this board tracks. 6. If the pool is met, wait or top up: Paid features pause until the weekly limit resets, while free-tier limits stay available. Extra Usage Credits are used only after the included weekly usage is exhausted, and top-up credits expire one year after purchase. 7. For Grok Bot, check Usage & Billing: Grok Bot has its own weekly usage, separate from your Grok and Cursor plans. Review it from the account access page or Usage & Billing. If both a Cursor and a SuperGrok subscription are linked, "Grok Bot uses whichever has more usage." FAQ: Q: Is there a /status or /usage command for Grok? A: The xAI documents linked below describe Settings → Usage for Grok and Usage & Billing for Grok Bot; neither page lists a chat command for checking usage. This site does not claim one exists or not beyond what those pages say. Q: How much weekly usage does SuperGrok include? A: Not published. x.ai/pricing describes SuperGrok Plus only as "Significantly higher usage across Chat, Imagine, Voice & Build", and the FAQ gives the reset schedule without a number. This site does not estimate it. Q: What does the Grok Bot limit notice look like? A: A Cursor user posted the banner on 2026-09-01: "You've reached your Grok Bot usage limit. It resets in 3 days." Cursor staff replied that "Your weekly usage resets on its own (the app shows the countdown)." Q: Has xAI ever reset everyone's weekly usage? A: Yes. On 2026-08-26, when Grok Bot was added to more plans, @bot posted "We're also resetting weekly usage limits for all users." That kind of announcement is what this board records as a reset. Sources: 1. xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) — https://docs.x.ai/grok/faq (verified 2026-09-17) 2. xAI docs: Grok Bot overview ("Usage resets weekly.") — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 3. xAI docs: Grok Bot FAQ (weekly usage, on-demand usage) — https://docs.x.ai/grok-bot/faq (verified 2026-09-17) 4. xAI docs: Grok Bot troubleshooting (Usage & Billing) — https://docs.x.ai/grok-bot/troubleshooting (verified 2026-09-17) 5. xAI news, 2026-08-26: Grok Bot is now included with more plans — https://x.ai/news/grok-bot-more-plans (verified 2026-09-17) 6. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 7. Cursor forum, 2026-09-01: Grok Bot usage-limit banner and Cursor staff reply — https://forum.cursor.com/t/grok-bot-weekly-usage-hits-100-after-bot-to-bot-reviews-the-user-asked-to-stop/170271 (verified 2026-09-17) 8. xAI pricing: SuperGrok plans (x.ai/pricing) — https://x.ai/pricing (verified 2026-09-17) ### 怎么在 Settings → Usage 查 Grok 剩余额度 URL: https://didreset.com/zh/guides/check-grok-usage Language: zh · Kind: 操作 · Published: 2026-09-18 · Updated: 2026-09-18 打开 Grok 网页版或 App 的 Settings → Usage。xAI 的 FAQ 写明，这个标签页有显示当前用量百分比的进度条、按产品划分的百分比，以及周用量池的刷新时间表。Grok Bot 的用量与账单，xAI 的故障排查页指向账户访问页里的 Usage & Billing。xAI 没有公布任何计划的具体额度数字，本站也不估算。 定义 付费 Grok 计划从一个跨产品共享的周用量池计量；xAI 原话："Different products cost different amounts depending on how much compute that product requires."。Settings 里的 Usage 标签页就是查看这个池子、各产品占比和刷新时间表的地方。xAI 的 FAQ 只描述了这个周池，没有提到 5 小时窗口。 池子用完时："When your weekly usage limit is met, paid features will pause until your weekly limit resets. You will still have access to Grok's free tier limits."。Extra Usage Credits "are used after your plan's included weekly usage is exhausted. Top up credits expire one year after purchase." Grok Bot 单独计量。xAI 的新闻稿说 "Grok Bot comes with its own usage, separate from your Grok and Cursor plans"，概览页写 "Usage resets weekly."。用量耗尽或达到按需消费上限时，故障排查页说 "review usage and billing from the account access page or Usage & Billing when that section is available." 官方文档怎么说 - "A progress bar showing your current usage percentage" — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) - "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) - "When your weekly usage limit is met, paid features will pause until your weekly limit resets." — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) - "Usage resets weekly." — xAI docs: Grok Bot overview ("Usage resets weekly.") (https://docs.x.ai/grok-bot/overview) - "review usage and billing from the account access page or Usage & Billing when that section is available" — xAI docs: Grok Bot troubleshooting (Usage & Billing) (https://docs.x.ai/grok-bot/troubleshooting) 步骤 1. 打开 Grok 的 Settings: 在网页版或 App 登录 Grok，打开 Settings。用量视图是 Settings 里的一个标签页，不是聊天命令。 2. 打开 Usage 标签页: FAQ 把 Settings 里的 Usage 标签页描述为显示周用量池和刷新时间表的地方。 3. 读进度条: 这条进度条是 "A progress bar showing your current usage percentage"，表示周池已用百分比。xAI 没有公布任何计划的 100% 对应多少。 4. 读按产品的占比: 进度条下面是 "A percentage breakdown by product"，能看出是 Chat、Imagine、Voice 还是 Build 把池子用掉的。不同产品按所需算力计费不同。 5. 记下周刷新时间表: "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings."。这个时间表是你自己账号的时钟，不是本站出发板追踪的全员重置。 6. 池子用完：等待或加购: 付费功能暂停到周限额刷新为止，免费档的额度仍然可用。Extra Usage Credits 只在计划内含的周用量耗尽后才开始扣，充值的积分自购买起一年后过期。 7. Grok Bot 看 Usage & Billing: Grok Bot 有自己独立的周用量，与 Grok 和 Cursor 计划分开。到账户访问页或 Usage & Billing 查看。如果同时绑定了 Cursor 和 SuperGrok 订阅，"Grok Bot uses whichever has more usage." FAQ: Q: Grok 有 /status 或 /usage 命令吗？ A: 下方链接的 xAI 文档只描述了 Grok 的 Settings → Usage 和 Grok Bot 的 Usage & Billing，两页都没有列出查额度的聊天命令。本站不在这些页面之外做有或没有的断言。 Q: SuperGrok 每周包含多少用量？ A: 未公布。x.ai/pricing 对 SuperGrok Plus 只写 "Significantly higher usage across Chat, Imagine, Voice & Build"，FAQ 给了刷新时间表但没有数字。本站不估算。 Q: Grok Bot 到限额时的提示长什么样？ A: 一位 Cursor 用户在 2026-09-01 贴出横幅原文："You've reached your Grok Bot usage limit. It resets in 3 days."。Cursor 员工回复 "Your weekly usage resets on its own (the app shows the countdown)." Q: xAI 有没有给所有人重置过周用量？ A: 有。2026-08-26 Grok Bot 扩展到更多计划时，@bot 发帖 "We're also resetting weekly usage limits for all users."。这类公告就是本站出发板记录的重置。 Sources: 1. xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) — https://docs.x.ai/grok/faq (verified 2026-09-17) 2. xAI docs: Grok Bot overview ("Usage resets weekly.") — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 3. xAI docs: Grok Bot FAQ (weekly usage, on-demand usage) — https://docs.x.ai/grok-bot/faq (verified 2026-09-17) 4. xAI docs: Grok Bot troubleshooting (Usage & Billing) — https://docs.x.ai/grok-bot/troubleshooting (verified 2026-09-17) 5. xAI news, 2026-08-26: Grok Bot is now included with more plans — https://x.ai/news/grok-bot-more-plans (verified 2026-09-17) 6. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 7. Cursor forum, 2026-09-01: Grok Bot usage-limit banner and Cursor staff reply — https://forum.cursor.com/t/grok-bot-weekly-usage-hits-100-after-bot-to-bot-reviews-the-user-asked-to-stop/170271 (verified 2026-09-17) 8. xAI pricing: SuperGrok plans (x.ai/pricing) — https://x.ai/pricing (verified 2026-09-17) ### How to redeem a Codex banked reset with /usage or Settings → Usage URL: https://didreset.com/guides/redeem-codex-banked-reset Language: en · Kind: How-to · Published: 2026-09-18 · Updated: 2026-09-18 Open Settings → Usage in the ChatGPT desktop app, on web or on mobile, or type /usage in the Codex CLI and choose to redeem an earned reset. Select the usage summary showing "1 reset available" or "Full reset", review the expiration, then confirm. OpenAI's help center says a full banked reset "refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." Banked resets are usable for 30 days after they are granted. Definition A banked reset is "a one-time Codex usage-limit reset saved to your account until you use it or it expires. It is not purchased credit and does not permanently increase your plan's usage limit." OpenAI Support adds that banked resets "aren't something you can normally purchase"; they arrive through promotions such as launch rewards or referrals. Redemption is manual. A banked reset "must be applied by you", whereas an automatic or global reset "won't appear as a saved reset in Settings." The reset "is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available." Timing matters. The pricing page says "Banked rate-limit resets are usable for 30 days after they're granted", and the help center warns "Once an unused reset expires, it cannot be restored or reissued." Redeeming from web and mobile was added on 2026-07-12; before that only the desktop app could do it. What the official docs say - "A banked reset is a one-time Codex usage-limit reset saved to your account until you use it or it expires." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Type /usage to open the usage menu. Choose whether to show token activity or redeem an available earned reset." — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) - "open Settings → Usage in the ChatGPT desktop app and select an option such as "1 reset available" or "Full reset."" — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Banked rate-limit resets are usable for 30 days after they're granted." — Codex pricing and usage limits (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Once an unused reset expires, it cannot be restored or reissued." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) Steps 1. Confirm you have one: The usage summary in Settings → Usage shows "1 reset available" or "Full reset" when a banked reset is waiting. In the desktop app a banner may read "You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days." 2. Decide whether to use it now: A reset is consumed only when it refreshes at least one eligible window; if there is nothing to reset, it stays available. But it expires: "Check the expiration shown in your account or offer." 3. Open the redeem menu: Desktop app, web or mobile: open Settings → Usage. Codex CLI: type /usage to open the usage menu and choose to redeem an available earned reset. OpenAI's wording: "Open settings → usage on the desktop app, CLI, or on web". 4. Select the reset: Pick the usage summary showing an available reset, such as "1 reset available" or "Full reset." 5. Review the expiration, then confirm: "Review the expiration details, then confirm that you want to use it." A full reset refreshes both the 5-hour and weekly windows at once. 6. Verify in Settings → Usage: "Check Settings → Usage for your updated reset time." The dashboard "may take a short time to update. If the percentage does not change immediately, refresh or reopen Settings → Usage". 7. Note your new weekly date: A full banked reset "changes your weekly reset date." Tibo Sottiaux on 2026-07-13: "You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you." FAQ: Q: Can I buy a banked reset? A: No. OpenAI Support: "Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase." A purchased instant reset is a different product that applies at checkout and "cannot be saved for later." Q: Does an official global reset use up my banked reset? A: No. "An automatic or global reset refreshes the eligible usage limits specified in the announcement. You don't need to apply it yourself, and it won't appear as a saved reset in Settings." Your banked reset stays until you use it or it expires. Q: Can I hold more than one banked reset? A: Yes. Tibo Sottiaux noted on 2026-06-28 that "some users had stacked up to three banked resets already that they can apply on their own schedule." Each has its own expiration shown in your account. Q: A reset was announced but nothing shows in my account. What now? A: Eligibility and any qualifying cutoff depend on the specific offer. OpenAI says support can review eligibility, but "Reporting an issue alone does not qualify you for an additional reset or compensation", and "Individual access issues after launch do not qualify you for an additional reset." Sources: 1. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) 4. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 5. OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) 6. Tibo Sottiaux on X, 2026-07-12: banked reset redeemable from web and mobile — https://x.com/thsottiaux/status/2076418567143408112 (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-07-13: applying a banked reset replenishes weekly usage — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-06-28: users stacked up to three banked resets — https://x.com/thsottiaux/status/2071381664853319742 (verified 2026-09-17) 9. openai/codex issue #37396: "You have a new rate limit reset available" banner (2026-08-07) — https://github.com/openai/codex/issues/37396 (verified 2026-09-17) ### 怎么用 /usage 或 Settings → Usage 兑换 Codex 重置卡 URL: https://didreset.com/zh/guides/redeem-codex-banked-reset Language: zh · Kind: 操作 · Published: 2026-09-18 · Updated: 2026-09-18 在 ChatGPT 桌面 App、网页或手机端打开 Settings → Usage，或在 Codex CLI 输入 /usage 选择兑换已获得的重置。点「1 reset available」或「Full reset」，看清过期时间再确认。OpenAI 帮助中心写明，完整重置卡同时刷新 5 小时与周窗口并改变周刷新日期；自发放起 30 天内可用。 定义 重置卡（banked reset）是 "a one-time Codex usage-limit reset saved to your account until you use it or it expires. It is not purchased credit and does not permanently increase your plan's usage limit."。OpenAI Support 补充，重置卡 "aren't something you can normally purchase"，它来自发布奖励、推荐邀请之类的促销活动。 兑换是手动的。重置卡 "must be applied by you"，而自动或全局重置 "won't appear as a saved reset in Settings."。重置卡 "is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available." 时间很关键。定价页写 "Banked rate-limit resets are usable for 30 days after they're granted"，帮助中心警告 "Once an unused reset expires, it cannot be restored or reissued."。网页和手机端兑换是 2026-07-12 加上的，此前只能在桌面 App 兑换。 官方文档怎么说 - "A banked reset is a one-time Codex usage-limit reset saved to your account until you use it or it expires." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Type /usage to open the usage menu. Choose whether to show token activity or redeem an available earned reset." — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) - "open Settings → Usage in the ChatGPT desktop app and select an option such as "1 reset available" or "Full reset."" — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Banked rate-limit resets are usable for 30 days after they're granted." — Codex pricing and usage limits (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Once an unused reset expires, it cannot be restored or reissued." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) 步骤 1. 确认你手里有一张: 有重置卡待用时，Settings → Usage 的用量摘要会显示「1 reset available」或「Full reset」。桌面 App 可能弹出横幅 "You have a new rate limit reset available — You were granted a rate limit reset that will expire in 30 days." 2. 决定现在用不用: 重置卡只有在成功刷新了至少一个符合条件的窗口时才会被消耗；没有可刷新的就继续留着。但它会过期："Check the expiration shown in your account or offer." 3. 打开兑换入口: 桌面 App、网页或手机：打开 Settings → Usage。Codex CLI：输入 /usage 打开用量菜单，选择兑换已获得的重置。OpenAI 原话："Open settings → usage on the desktop app, CLI, or on web"。 4. 选中那张重置卡: 点显示可用重置的用量摘要，例如「1 reset available」或「Full reset」。 5. 看过期信息，然后确认: "Review the expiration details, then confirm that you want to use it."。完整重置会同时刷新 5 小时窗口和周窗口。 6. 到 Settings → Usage 核对: "Check Settings → Usage for your updated reset time."。仪表盘 "may take a short time to update. If the percentage does not change immediately, refresh or reopen Settings → Usage"。 7. 记下新的周刷新日期: 完整重置卡 "changes your weekly reset date."。Tibo Sottiaux 2026-07-13："You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you." FAQ: Q: 重置卡能买吗？ A: 不能。OpenAI Support："Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase."。购买的即时重置是另一种产品，结账时立即生效，"cannot be saved for later." Q: 官方全员重置会消耗我的重置卡吗？ A: 不会。"An automatic or global reset refreshes the eligible usage limits specified in the announcement. You don't need to apply it yourself, and it won't appear as a saved reset in Settings."。你的重置卡一直留到用掉或过期。 Q: 能同时存多张吗？ A: 能。Tibo Sottiaux 在 2026-06-28 提到 "some users had stacked up to three banked resets already that they can apply on their own schedule."。每张的过期时间在账号里各自显示。 Q: 公告发了重置，我的账号里却没有，怎么办？ A: 资格和截止时间取决于具体活动。OpenAI 说支持团队可以复核资格，但 "Reporting an issue alone does not qualify you for an additional reset or compensation"，并且 "Individual access issues after launch do not qualify you for an additional reset." Sources: 1. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) 4. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 5. OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) 6. Tibo Sottiaux on X, 2026-07-12: banked reset redeemable from web and mobile — https://x.com/thsottiaux/status/2076418567143408112 (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-07-13: applying a banked reset replenishes weekly usage — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2026-06-28: users stacked up to three banked resets — https://x.com/thsottiaux/status/2071381664853319742 (verified 2026-09-17) 9. openai/codex issue #37396: "You have a new rate limit reset available" banner (2026-08-07) — https://github.com/openai/codex/issues/37396 (verified 2026-09-17) ### How to switch model or account after a usage limit with /model, /login and /logout URL: https://didreset.com/guides/switch-model-or-account-when-limited Language: en · Kind: How-to · Published: 2026-09-18 · Updated: 2026-09-18 No, switching models does not restore a shared pool. OpenAI: "Switching models does not restore allowance in a shared usage pool." Anthropic's docs say a subscription window is "shared across all models, so the developer can't restore access by switching models with /model"; only a model-family limit such as "You've hit your Opus limit" can be sidestepped by switching families. To change accounts, use /logout in Codex or /login in Claude Code. Definition Both vendors meter subscription usage as one pool per account. OpenAI: "Depending on your account, you may be able to wait for the reset, use a saved reset, or continue with eligible credits." Claude Code offers the same shape of choice: wait for the reset (since v2.1.234 it can wait and continue the interrupted task automatically via /rate-limit-options) or enable pay-as-you-go with /usage-credits. The exception is a model-family limit in Claude Code. The error reference lists "You've hit your Opus limit" and "You've hit your Sonnet limit"; those apply only to requests to that family, and "switching to a model outside that family with /model does keep the developer working." Each model has its own prompt cache, so the next request re-reads the whole conversation with no cache hits. Changing accounts is a separate action from changing models. Codex CLI: "/logout — Sign out of Codex. Clear local credentials when using a shared machine." Claude Code: "run /login from within Claude Code to switch to your subscription plan", or log out fully and select the correct account. xAI's FAQ describes no account switching; it only says "When your weekly usage limit is met, paid features will pause until your weekly limit resets." What the official docs say - "Switching models does not restore allowance in a shared usage pool." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "shared across all models, so the developer can't restore access by switching models with /model" — Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) (https://code.claude.com/docs/en/costs) - "switching to a model outside that family with /model does keep the developer working." — Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) (https://code.claude.com/docs/en/costs) - "/logout — Sign out of Codex. Clear local credentials when using a shared machine." — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) - "run /login from within Claude Code to switch to your subscription plan" — Use Claude Code with your Pro or Max plan (Claude Help Center) (https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) - "When your weekly usage limit is met, paid features will pause until your weekly limit resets." — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) Steps 1. Identify which limit you hit: Codex: type /status and read which bar is empty. Claude Code: type /usage, or read the message itself. "You've hit your session limit" and "You've hit your weekly limit" are shared-pool limits; "You've hit your Opus limit" and "You've hit your Sonnet limit" are family limits. 2. If it is a shared pool, do not bother switching models: OpenAI: "Switching models does not restore allowance in a shared usage pool." Anthropic: the window is "shared across all models, so the developer can't restore access by switching models with /model". Your options are to wait, use a saved reset (Codex), or use credits. 3. If it is a Claude Code family limit, use /model: Switch to a model outside the Opus or Sonnet family and keep working. Expect the next request to re-read the whole conversation, because each model has its own prompt cache. 4. Codex: change account with /logout: "/logout — Sign out of Codex. Clear local credentials when using a shared machine." Then sign in again with the other account. The other account has its own 5-hour and weekly windows. 5. Claude Code: change account with /login: Anthropic: "run /login from within Claude Code to switch to your subscription plan". If the wrong account is active, log out fully and select the correct account when you sign back in. 6. Or wait in place: Claude Code v2.1.234 and later can wait and continue the interrupted task automatically after the reset; the waiting line reads "Usage limit reached · continuing automatically at 3:45pm · esc to cancel". Codex: "you may be able to wait for the reset". 7. Check the board before you switch anything: If an account-wide reset has just been announced, the board shows RESET NOW and every account is refilled. Your own window is separate from that, but a global reset makes switching unnecessary. FAQ: Q: Does a cheaper model make the pool last longer before I hit the limit? A: Model choice affects how fast the pool drains, not whether switching refills it. OpenAI: "Model choice, context, reasoning, tool use, retrieval, and caching all affect usage". Claude Code's docs suggest "model: haiku" for simple subagent tasks. Once the pool is empty, switching does not restore it. Q: Which Claude Code limits can /model get around? A: Only the Opus and Sonnet family limits. "The Opus and Sonnet limits each apply only to requests to that model family"; session and weekly limits are shared across all models. Q: What about Grok? A: xAI's FAQ: "When your weekly usage limit is met, paid features will pause until your weekly limit resets. You will still have access to Grok's free tier limits." Extra Usage Credits are used after the included weekly usage is exhausted. Q: Will I lose my task if I wait? A: In Claude Code v2.1.234+ the session waits and continues automatically; the docs show "Usage limit reached · continuing automatically at 3:45pm · esc to cancel" and "Usage limit reset · continuing automatically". Codex's options are wait, saved reset, or credits. Sources: 1. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) — https://code.claude.com/docs/en/costs (verified 2026-09-17) 3. Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) 4. Use Claude Code with your Pro or Max plan (Claude Help Center) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 5. xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) — https://docs.x.ai/grok/faq (verified 2026-09-17) 6. Claude Code docs: Error reference (session, weekly, Opus and Sonnet limits) — https://code.claude.com/docs/en/errors (verified 2026-09-17) 7. Claude Code docs: Interactive mode (waiting for a usage limit to reset) — https://code.claude.com/docs/en/interactive-mode (verified 2026-09-17) 8. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) ### 被限额后怎么用 /model、/login、/logout 切模型或切账号 URL: https://didreset.com/zh/guides/switch-model-or-account-when-limited Language: zh · Kind: 操作 · Published: 2026-09-18 · Updated: 2026-09-18 不行，换模型不会让共享池回血。OpenAI 写明，在共享用量池里切换模型不会恢复额度；Anthropic 的文档也说订阅窗口跨所有模型共享，不能靠 /model 恢复访问。只有按模型家族计的 Opus 或 Sonnet 限额，换到家族外的模型才能继续。换账号：Codex 用 /logout，Claude Code 用 /login。 定义 两家都按账号把订阅用量计成一个池子。OpenAI："Depending on your account, you may be able to wait for the reset, use a saved reset, or continue with eligible credits."。Claude Code 的选项形状一样：等刷新（v2.1.234 起可通过 /rate-limit-options 原地等待并自动续跑被打断的任务），或用 /usage-credits 开启按量付费。 例外是 Claude Code 的模型家族限额。错误参考列出 "You've hit your Opus limit" 和 "You've hit your Sonnet limit"，它们只作用于对该家族的请求，"switching to a model outside that family with /model does keep the developer working."。每个模型有自己的 prompt cache，所以下一次请求会重读整个对话，没有缓存命中。 换账号和换模型是两件事。Codex CLI："/logout — Sign out of Codex. Clear local credentials when using a shared machine."。Claude Code："run /login from within Claude Code to switch to your subscription plan"，或者完全登出后选择正确的账号。xAI 的 FAQ 没有描述换账号，只说 "When your weekly usage limit is met, paid features will pause until your weekly limit resets." 官方文档怎么说 - "Switching models does not restore allowance in a shared usage pool." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "shared across all models, so the developer can't restore access by switching models with /model" — Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) (https://code.claude.com/docs/en/costs) - "switching to a model outside that family with /model does keep the developer working." — Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) (https://code.claude.com/docs/en/costs) - "/logout — Sign out of Codex. Clear local credentials when using a shared machine." — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) - "run /login from within Claude Code to switch to your subscription plan" — Use Claude Code with your Pro or Max plan (Claude Help Center) (https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) - "When your weekly usage limit is met, paid features will pause until your weekly limit resets." — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) 步骤 1. 先弄清撞的是哪个限额: Codex：输入 /status 看哪条进度条空了。Claude Code：输入 /usage，或直接读报错。"You've hit your session limit" 和 "You've hit your weekly limit" 是共享池限额；"You've hit your Opus limit" 和 "You've hit your Sonnet limit" 是家族限额。 2. 共享池满了，别费劲换模型: OpenAI："Switching models does not restore allowance in a shared usage pool."。Anthropic：窗口 "shared across all models, so the developer can't restore access by switching models with /model"。你的选项是等待、用重置卡（Codex）或用积分。 3. Claude Code 家族限额：用 /model: 切到 Opus 或 Sonnet 家族之外的模型继续工作。下一次请求会重读整个对话，因为每个模型有自己的 prompt cache。 4. Codex 换账号：/logout: "/logout — Sign out of Codex. Clear local credentials when using a shared machine."。然后用另一个账号重新登录。另一个账号有它自己的 5 小时窗口和周窗口。 5. Claude Code 换账号：/login: Anthropic："run /login from within Claude Code to switch to your subscription plan"。如果当前是错误账号，完全登出后在重新登录时选择正确的账号。 6. 或者原地等: Claude Code v2.1.234 及以后可以原地等刷新并自动续跑被打断的任务，等待行文案是 "Usage limit reached · continuing automatically at 3:45pm · esc to cancel"。Codex："you may be able to wait for the reset"。 7. 动手之前先看一眼出发板: 如果刚刚公告了全员重置，出发板显示「已重置」，每个账号都被刷新了。你自己的窗口和它是两回事，但全员重置意味着不用切换。 FAQ: Q: 用便宜的模型能让池子撑得更久吗？ A: 模型选择影响池子消耗的速度，不影响换了之后会不会回血。OpenAI："Model choice, context, reasoning, tool use, retrieval, and caching all affect usage"。Claude Code 文档建议简单的 subagent 任务用 "model: haiku"。池子一旦空了，换模型不会恢复。 Q: Claude Code 的哪些限额能用 /model 绕过？ A: 只有 Opus 和 Sonnet 家族限额。"The Opus and Sonnet limits each apply only to requests to that model family"；会话限额和周限额跨所有模型共享。 Q: Grok 呢？ A: xAI 的 FAQ："When your weekly usage limit is met, paid features will pause until your weekly limit resets. You will still have access to Grok's free tier limits."。Extra Usage Credits 在计划内含的周用量耗尽后才开始扣。 Q: 等待期间任务会丢吗？ A: Claude Code v2.1.234+ 会话会原地等待并自动继续，文档里的文案是 "Usage limit reached · continuing automatically at 3:45pm · esc to cancel" 和 "Usage limit reset · continuing automatically"。Codex 的选项是等待、重置卡或积分。 Sources: 1. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Claude Code docs: Manage costs effectively (/usage, /model, /rate-limit-options) — https://code.claude.com/docs/en/costs (verified 2026-09-17) 3. Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) 4. Use Claude Code with your Pro or Max plan (Claude Help Center) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 5. xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) — https://docs.x.ai/grok/faq (verified 2026-09-17) 6. Claude Code docs: Error reference (session, weekly, Opus and Sonnet limits) — https://code.claude.com/docs/en/errors (verified 2026-09-17) 7. Claude Code docs: Interactive mode (waiting for a usage limit to reset) — https://code.claude.com/docs/en/interactive-mode (verified 2026-09-17) 8. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) ### How to buy a Codex usage reset URL: https://didreset.com/guides/buy-codex-usage-reset Language: en · Kind: How-to · Published: 2026-09-18 · Updated: 2026-09-18 Yes, but only an instant reset, and only for some accounts. OpenAI's help center: "Purchased instant resets are available to eligible personal ChatGPT Plus and Pro accounts, depending on your account and billing country." It "refreshes your five-hour and weekly Work and Codex allowance as soon as checkout succeeds, even if you still have usage remaining. It cannot be saved for later." Banked resets are promotional and not normally purchasable. OpenAI has not published a price. Definition Three different things can restore Codex allowance: the windows rolling over on their own, a banked reset you redeem, and a purchased instant reset. Only the last one costs money, and it is not credit: "Credits pay for eligible additional usage under their own terms. A reset is not cash or API credit." A purchased reset moves your weekly clock. "Purchasing a reset brings your weekly allowance forward. Your new weekly period starts with your first Work or Codex request after the reset", and your next automatic weekly reset is seven days after that request. Availability is documented in the help-center revision dated 2026-09-16; the page does not state a price, and this site does not guess one. OpenAI Support says banked resets "aren't something you can normally purchase"; eligible Plus and Pro users "may receive them through limited promotions, such as launch rewards or successful referrals." What the official docs say - "Purchased instant resets are available to eligible personal ChatGPT Plus and Pro accounts, depending on your account and billing country." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "It cannot be saved for later." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Purchasing a reset brings your weekly allowance forward." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase." — OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) (https://community.openai.com/t/how-banked-reset-works-in-codex/1388104) - "Eligible Plus and Pro users can buy credits without changing plans." — Using Codex with your ChatGPT plan (OpenAI Help Center) (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) Steps 1. See what is exhausted: Type /status in the Codex CLI or open Settings → Usage. OpenAI: "Check which allowance is exhausted, your credit balance, and any reset time shown." 2. Use a banked reset first, if you have one: A banked reset is free and expires 30 days after it is granted, while a purchased reset "cannot be saved for later." If Settings → Usage shows "1 reset available", redeem that before paying. 3. Confirm you are eligible: Purchased instant resets are for "eligible personal ChatGPT Plus and Pro accounts, depending on your account and billing country." The help center names no other plan. 4. Weigh the alternatives: "You may be able to add credits, use an available reset, upgrade, or wait until the displayed reset time." Credits are a separate product with their own terms; "Eligible Plus and Pro users can buy credits without changing plans." 5. Complete checkout: The reset "refreshes your five-hour and weekly Work and Codex allowance as soon as checkout succeeds, even if you still have usage remaining." The price is not published on the help-center page. 6. Note the new weekly clock: "Your new weekly period starts with your first Work or Codex request after the reset", and the next automatic weekly reset lands seven days after that request. "Check Settings → Usage for your updated reset time." FAQ: Q: How much does a purchased reset cost? A: Not published. The help-center article that documents purchased instant resets (revision dated 2026-09-16) gives eligibility and effect but no price. This site does not estimate it. Q: Can Business or Enterprise seats buy one? A: The help center names only "eligible personal ChatGPT Plus and Pro accounts". It says nothing about Business or Enterprise, so this site does not claim either way. Q: Does a purchased reset also use up my banked reset? A: No. A banked reset "is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available." The purchased reset applies at checkout on its own. Q: Is a purchased reset the same as a global reset on the board? A: No. A global reset "refreshes the eligible usage limits specified in the announcement" for everyone and "won't appear as a saved reset in Settings"; the one documented for 2026-09-07 covered Plus, Pro and Business. A purchased reset is per account, at checkout. Sources: 1. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) 3. Using Codex with your ChatGPT plan (OpenAI Help Center) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 4. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 5. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) ### 怎么购买 Codex 即时重置 URL: https://didreset.com/zh/guides/buy-codex-usage-reset Language: zh · Kind: 操作 · Published: 2026-09-18 · Updated: 2026-09-18 可以，但只能买即时重置，而且只对部分账号开放。OpenAI 帮助中心写明：购买的即时重置面向符合条件的个人 ChatGPT Plus 与 Pro 账号，取决于账号与账单国家；结账成功后立刻刷新 5 小时与周额度，即使你还有剩余用量也会刷新，不能存起来以后用。重置卡是促销发放的，通常不能购买。OpenAI 没有公布价格，本站不估算。 定义 能恢复 Codex 额度的有三样东西：窗口自己到点刷新、你手动兑换的重置卡、花钱买的即时重置。只有最后一种要花钱，而且它不是积分："Credits pay for eligible additional usage under their own terms. A reset is not cash or API credit." 购买的重置会移动你的周时钟。"Purchasing a reset brings your weekly allowance forward. Your new weekly period starts with your first Work or Codex request after the reset"，下一次自动周刷新在那次请求的七天之后。 这项功能记载于 2026-09-16 版本的帮助中心文章；页面没有写价格，本站也不猜。OpenAI Support 说重置卡 "aren't something you can normally purchase"，符合条件的 Plus 与 Pro 用户 "may receive them through limited promotions, such as launch rewards or successful referrals." 官方文档怎么说 - "Purchased instant resets are available to eligible personal ChatGPT Plus and Pro accounts, depending on your account and billing country." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "It cannot be saved for later." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Purchasing a reset brings your weekly allowance forward." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase." — OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) (https://community.openai.com/t/how-banked-reset-works-in-codex/1388104) - "Eligible Plus and Pro users can buy credits without changing plans." — Using Codex with your ChatGPT plan (OpenAI Help Center) (https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan) 步骤 1. 先看哪个额度用完了: 在 Codex CLI 输入 /status，或打开 Settings → Usage。OpenAI："Check which allowance is exhausted, your credit balance, and any reset time shown." 2. 有重置卡先用重置卡: 重置卡免费，自发放起 30 天过期；而购买的重置 "cannot be saved for later."。如果 Settings → Usage 显示「1 reset available」，先兑换它再考虑花钱。 3. 确认你符合条件: 购买的即时重置面向 "eligible personal ChatGPT Plus and Pro accounts, depending on your account and billing country."。帮助中心没有点名其他计划。 4. 权衡其他选项: "You may be able to add credits, use an available reset, upgrade, or wait until the displayed reset time."。积分是另一种产品，有自己的条款；"Eligible Plus and Pro users can buy credits without changing plans." 5. 完成结账: 重置 "refreshes your five-hour and weekly Work and Codex allowance as soon as checkout succeeds, even if you still have usage remaining."。帮助中心页面上没有公布价格。 6. 记下新的周时钟: "Your new weekly period starts with your first Work or Codex request after the reset"，下一次自动周刷新在那次请求的七天之后。"Check Settings → Usage for your updated reset time." FAQ: Q: 购买的重置多少钱？ A: 未公布。记载购买即时重置的帮助中心文章（2026-09-16 版本）只写了资格和效果，没有价格。本站不估算。 Q: Business 或 Enterprise 座席能买吗？ A: 帮助中心只点名 "eligible personal ChatGPT Plus and Pro accounts"，对 Business 和 Enterprise 只字未提，所以本站不做任何一方向的断言。 Q: 购买的重置会顺带消耗我的重置卡吗？ A: 不会。重置卡 "is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available."。购买的重置在结账时独立生效。 Q: 购买的重置和出发板上的全员重置一样吗？ A: 不一样。全员重置 "refreshes the eligible usage limits specified in the announcement"，对所有人生效且 "won't appear as a saved reset in Settings"；帮助中心记载的 2026-09-07 那次覆盖 Plus、Pro 和 Business。购买的重置只作用于你的账号，在结账时生效。 Sources: 1. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) 3. Using Codex with your ChatGPT plan (OpenAI Help Center) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 4. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 5. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) ### What is a reset? URL: https://didreset.com/guides/reset Language: en · Kind: Term · Term code: reset · Published: 2026-09-18 · Updated: 2026-09-18 A reset, on this board, is an account-wide usage-limit reset that a vendor announces and applies to everyone at once. OpenAI describes it: "An automatic or global reset refreshes the eligible usage limits specified in the announcement. You don't need to apply it yourself, and it won't appear as a saved reset in Settings." One documented Codex example is the global reset of 2026-09-07. It is not your own window refreshing. Definition The word "reset" gets used for three different things. Your 5-hour window and weekly limit refresh on their own clock, per account; that is a window roll-over, and the time is shown by /status (Codex), /usage (Claude Code) or Settings → Usage (Grok). A banked reset is a saved card you redeem yourself. A reset in the sense this site tracks is a public announcement that the vendor has refreshed limits for a whole population of accounts. Announcements come from named official accounts or vendor help pages, and every row on the board links to the original. Two examples: OpenAI's help center documents a global automatic reset of Plus, Pro and Business limits on 2026-09-07, applied immediately and not saved as a banked reset; on 2026-08-26 Grok Bot's @bot account posted "We're also resetting weekly usage limits for all users." A global reset does not change how the windows behave afterwards. OpenAI notes that after a reset "you can reach the five-hour limit while you still have weekly usage remaining." A banked reset you were holding is untouched: a global reset "won't appear as a saved reset in Settings" and does not consume the one already there. On the board, a reset announced within the last 24 hours shows as RESET NOW; afterwards the row returns to WAITING. LAST RESET is the most recent announcement of this kind, and NEXT EST. adds the median gap of the last 10 resets to it: a cadence estimate, not a promise. OVERDUE means the wait has passed 1.5 times that median. BANKED and TEASE describe other signals, not resets. These rules are the site's own and are exposed in the public status API. What the official docs say - "An automatic or global reset refreshes the eligible usage limits specified in the announcement." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You don't need to apply it yourself, and it won't appear as a saved reset in Settings." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "We're also resetting weekly usage limits for all users." — Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users (https://x.com/bot/status/2092664183334846840) - "you can reach the five-hour limit while you still have weekly usage remaining." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) FAQ: Q: Is a reset the same as my 5-hour window refreshing? A: No. Your window refreshes on your own clock, per account, whether or not anyone announces anything; Codex shows that time in /status and Claude Code in /usage. A reset on this board is an announcement that limits were refreshed for everyone, and it is dated by the announcement. Q: Who announces resets? A: The official accounts this board watches: @thsottiaux for Codex, @ClaudeDevs and @lydiahallie for Claude, and @grok, @bot and @elonmusk for Grok, plus vendor help pages when they document one. Each row shows the announcing account and links to the post; the public status API returns the same fields. Q: Are resets on a schedule? A: No. None of the three vendors announces resets in advance. NEXT EST. is the last reset plus the median gap of the last 10 resets on record; the limits page and the time-zone pages also print the real range of recent gaps, which is wide. Q: Does a global reset use up my banked reset? A: No. "An automatic or global reset refreshes the eligible usage limits specified in the announcement. You don't need to apply it yourself, and it won't appear as a saved reset in Settings." A banked reset is consumed only when you apply it and it refreshes a window. Sources: 1. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 3. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-07-13: applying a banked reset replenishes weekly usage — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) 5. DidReset public status API: board status and cadence fields — https://didreset.com/api/status (verified 2026-09-17) 6. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 7. What is the Max plan? (Claude Help Center) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) ### 什么是重置？ URL: https://didreset.com/zh/guides/reset Language: zh · Kind: 术语 · Term code: reset · Published: 2026-09-18 · Updated: 2026-09-18 本站出发板上的「重置」，指厂商公告并对所有账号同时生效的额度重置。OpenAI 帮助中心的描述是：自动或全局重置会刷新公告里指定的额度，你不需要自己去应用，它也不会以已存重置的形式出现在设置里。帮助中心记载的一个例子是 Codex 2026-09-07 的全局重置。它不是你自己的窗口到点刷新，也不是重置卡。 定义 「重置」这个词被用来指三件不同的事。你的 5 小时窗口和周限额按你账号自己的时钟刷新，这是窗口到点，时间在 /status（Codex）、/usage（Claude Code）或 Settings → Usage（Grok）里看。重置卡是存在账号里、要你自己兑换的一张卡。而本站追踪的「重置」，是厂商公开宣布已为一大批账号刷新额度的公告。 公告来自具名的官方账号或厂商帮助页，出发板上每一行都链接到原文。两个例子：OpenAI 帮助中心记载 2026-09-07 对 Plus、Pro、Business 额度做了一次全局自动重置，立即生效、不存为重置卡；2026-08-26 Grok Bot 的 @bot 账号发帖 "We're also resetting weekly usage limits for all users." 全局重置不会改变之后窗口的行为。OpenAI 说明，重置之后 "you can reach the five-hour limit while you still have weekly usage remaining."。你手里的重置卡也不受影响：全局重置 "won't appear as a saved reset in Settings"，也不会消耗已有的那张。 在出发板上，24 小时内公告的重置显示为「已重置」（RESET NOW）；之后这一行回到「等待中」（WAITING）。「上次重置」是最近一次这类公告，「预计下次」在它之上加近 10 次重置的中位间隔：这是按规律估计，不是承诺。「已超期」（OVERDUE）表示等待时间已超过该中位数的 1.5 倍。「重置卡」（BANKED）和「预告中」（TEASE）描述的是其他信号，不是重置。这些规则是本站自己的，公开状态 API 里能看到同样的字段。 官方文档怎么说 - "An automatic or global reset refreshes the eligible usage limits specified in the announcement." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You don't need to apply it yourself, and it won't appear as a saved reset in Settings." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "We're also resetting weekly usage limits for all users." — Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users (https://x.com/bot/status/2092664183334846840) - "you can reach the five-hour limit while you still have weekly usage remaining." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) FAQ: Q: 重置和我自己的 5 小时窗口刷新是一回事吗？ A: 不是。你的窗口按账号自己的时钟刷新，不管有没有人发公告；Codex 在 /status、Claude Code 在 /usage 里显示这个时间。本站说的重置是「所有人的额度被刷新了」的公告，日期以公告为准。 Q: 重置是谁公告的？ A: 本站盯着的官方账号：Codex 的 @thsottiaux，Claude 的 @ClaudeDevs 与 @lydiahallie，Grok 的 @grok、@bot 与 @elonmusk，以及厂商帮助页记载的重置。每一行都标出公告账号并链接到原帖，公开状态 API 返回同样的字段。 Q: 重置有固定时间表吗？ A: 没有。三家厂商都不会提前宣布重置。「预计下次」是上次重置加上有记录的近 10 次重置的中位间隔；额度上限页和各时区页还会打印近期间隔的真实范围，跨度很大。 Q: 全局重置会消耗我的重置卡吗？ A: 不会。"An automatic or global reset refreshes the eligible usage limits specified in the announcement. You don't need to apply it yourself, and it won't appear as a saved reset in Settings."。重置卡只有在你主动应用并成功刷新了某个窗口时才被消耗。 Sources: 1. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 3. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-07-13: applying a banked reset replenishes weekly usage — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) 5. DidReset public status API: board status and cadence fields — https://didreset.com/api/status (verified 2026-09-17) 6. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 7. What is the Max plan? (Claude Help Center) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) ### What is a banked reset? URL: https://didreset.com/guides/banked-reset Language: en · Kind: Term · Term code: banked-reset · Published: 2026-09-18 · Updated: 2026-09-18 A banked reset is, in OpenAI's words, "a one-time Codex usage-limit reset saved to your account until you use it or it expires. It is not purchased credit and does not permanently increase your plan's usage limit." You redeem it yourself from Settings → Usage or with /usage in the CLI; a full one refreshes both the 5-hour and weekly windows and changes your weekly reset date. It is usable for 30 days after it is granted. Definition Reset banking was added on 2026-06-11: the Codex changelog lists "rate-limit reset banking for Plus and Pro users, including one free reset at launch and referral invitations for earning more during the current promotion." Banked resets are promotional; OpenAI Support says they "aren't something you can normally purchase." A banked reset "must be applied by you", and it "is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available." A global reset is the opposite: applied by OpenAI to everyone, and it "won't appear as a saved reset in Settings." A purchased instant reset applies at checkout and "cannot be saved for later." "Banked rate-limit resets are usable for 30 days after they're granted", and "Once an unused reset expires, it cannot be restored or reissued." More than one can sit in an account: on 2026-06-28 Tibo Sottiaux noted "some users had stacked up to three banked resets already". Redemption from web and mobile arrived on 2026-07-12; before that only the desktop app could redeem. On the board, a banked-reset drop is a signal of kind "banked": the STATUS flap reads BANKED for 72 hours after the announcement, and the alert channels say "banked reset card" rather than "reset announced". It does not move LAST RESET or NEXT EST., because those count only account-wide resets. Redeeming your own card is something you do in your account; the board cannot see it. What the official docs say - "A banked reset is a one-time Codex usage-limit reset saved to your account until you use it or it expires." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Banked rate-limit resets are usable for 30 days after they're granted." — Codex pricing and usage limits (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase." — OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) (https://community.openai.com/t/how-banked-reset-works-in-codex/1388104) - "/usage — View account token usage or use a rate-limit reset." — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) FAQ: Q: Do banked resets expire? A: Yes. The pricing page: "Banked rate-limit resets are usable for 30 days after they're granted." The help center adds that expiration can vary by offer, and "Once an unused reset expires, it cannot be restored or reissued." Check the expiration shown in your account. Q: Can I buy a banked reset? A: No. OpenAI Support: "Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase." What OpenAI does sell to eligible personal Plus and Pro accounts is an instant reset that applies at checkout and "cannot be saved for later." Q: Does using one change my weekly countdown? A: Yes. "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date. Check Settings → Usage for your updated reset time." Q: Do Claude Code or Grok have banked resets? A: As of 2026-09-17, only OpenAI has published a help-center article on banked resets. The Anthropic and xAI documents linked from the other glossary pages describe windows that refresh on their own and account-wide resets, not a saved reset you redeem. Sources: 1. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 4. OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) 5. ChatGPT & Codex changelog, 2026-06-11 entry: reset banking (learn.chatgpt.com) — https://learn.chatgpt.com/docs/changelog (verified 2026-09-17) 6. Tibo Sottiaux on X, 2026-06-28: users stacked up to three banked resets — https://x.com/thsottiaux/status/2071381664853319742 (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-07-12: banked reset redeemable from web and mobile — https://x.com/thsottiaux/status/2076418567143408112 (verified 2026-09-17) 8. Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) 9. DidReset fares: which alert channels fire on which signals — https://didreset.com/pricing (verified 2026-09-17) ### 什么是重置卡（banked reset）？ URL: https://didreset.com/zh/guides/banked-reset Language: zh · Kind: 术语 · Term code: banked-reset · Published: 2026-09-18 · Updated: 2026-09-18 重置卡（banked reset）按 OpenAI 帮助中心的定义，是存在你账号里、用掉或过期为止的一次性 Codex 额度重置；它不是购买的积分，也不会永久提高计划上限。你在 Settings → Usage 或用 CLI 的 /usage 自己兑换；完整的一张同时刷新 5 小时与周窗口，并改变周刷新日期。自发放起 30 天内可用。 定义 重置银行（reset banking）是 2026-06-11 加上的：Codex 更新日志写着 "rate-limit reset banking for Plus and Pro users, including one free reset at launch and referral invitations for earning more during the current promotion."。重置卡是促销性质；OpenAI Support 说它们 "aren't something you can normally purchase." 重置卡 "must be applied by you"，并且 "is consumed only when it successfully refreshes at least one eligible usage window. If there is nothing to reset, it remains available."。全局重置正相反：由 OpenAI 对所有人应用，"won't appear as a saved reset in Settings."。购买的即时重置在结账时生效，"cannot be saved for later." "Banked rate-limit resets are usable for 30 days after they're granted"，而且 "Once an unused reset expires, it cannot be restored or reissued."。一个账号可以存多张：2026-06-28 Tibo Sottiaux 提到 "some users had stacked up to three banked resets already"。网页和手机端兑换是 2026-07-12 加上的，此前只有桌面 App 能兑换。 在出发板上，发放重置卡是一种 kind 为 banked 的信号：公告后 72 小时内状态翻牌显示「重置卡」（BANKED），提醒渠道的文案是「发放重置卡」而不是「全员重置」。它不会移动「上次重置」和「预计下次」，因为那两格只统计全员重置。你自己兑换那张卡是账号里的动作，出发板看不见。 官方文档怎么说 - "A banked reset is a one-time Codex usage-limit reset saved to your account until you use it or it expires." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." — How banked Codex resets work (OpenAI Help Center) (https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work) - "Banked rate-limit resets are usable for 30 days after they're granted." — Codex pricing and usage limits (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase." — OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) (https://community.openai.com/t/how-banked-reset-works-in-codex/1388104) - "/usage — View account token usage or use a rate-limit reset." — Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) (https://learn.chatgpt.com/docs/developer-commands?surface=cli) FAQ: Q: 重置卡会过期吗？ A: 会。定价页："Banked rate-limit resets are usable for 30 days after they're granted."。帮助中心补充过期时间可能因活动而异，并且 "Once an unused reset expires, it cannot be restored or reissued."。以账号里显示的过期时间为准。 Q: 重置卡能买吗？ A: 不能。OpenAI Support："Banked resets are promotional Codex rate-limit resets, they aren't something you can normally purchase."。OpenAI 向符合条件的个人 Plus 与 Pro 账号出售的是即时重置，结账时生效，"cannot be saved for later." Q: 用了重置卡，周倒计时会变吗？ A: 会。"Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date. Check Settings → Usage for your updated reset time." Q: Claude Code 或 Grok 有重置卡吗？ A: 截至 2026-09-17，只有 OpenAI 发布过关于重置卡的帮助中心文章。其他术语页链接的 Anthropic 与 xAI 文档描述的是自动刷新的窗口和全员重置，没有需要你兑换的已存重置。 Sources: 1. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 4. OpenAI Developer Community: "How banked reset works in Codex?" (OpenAI_Support reply, 2026-08-01) — https://community.openai.com/t/how-banked-reset-works-in-codex/1388104 (verified 2026-09-17) 5. ChatGPT & Codex changelog, 2026-06-11 entry: reset banking (learn.chatgpt.com) — https://learn.chatgpt.com/docs/changelog (verified 2026-09-17) 6. Tibo Sottiaux on X, 2026-06-28: users stacked up to three banked resets — https://x.com/thsottiaux/status/2071381664853319742 (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-07-12: banked reset redeemable from web and mobile — https://x.com/thsottiaux/status/2076418567143408112 (verified 2026-09-17) 8. Codex CLI developer commands: /status, /usage, /logout (learn.chatgpt.com) — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) 9. DidReset fares: which alert channels fire on which signals — https://didreset.com/pricing (verified 2026-09-17) ### What is a 5-hour window? URL: https://didreset.com/guides/5-hour-window Language: en · Kind: Term · Term code: five-hour-window · Published: 2026-09-18 · Updated: 2026-09-18 A 5-hour window is the short-term usage allowance that Codex and Claude Code meter per account. Codex: "A new window starts when you send your first message in Work or Codex after the previous window ends." Claude: "Your session-based usage limit will reset every five hours." It sits under a separate weekly limit, refreshes on your own clock, and is shown by /status (Codex) or /usage (Claude Code). xAI's Grok FAQ describes a weekly pool only. Definition Codex: "Depending on your plan, usage limits may apply over a five-hour window and a weekly window. Where both apply, you need allowance remaining in both to continue". The window is not a fixed clock: it starts with your first message after the previous window ends, and "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." The pricing page publishes per-plan estimates, for example 5–45 GPT-6 Astra local messages per five-hour period on Plus, that "are not fixed message limits". Claude Pro and Max: "Your session-based usage limit will reset every five hours." Max 5x and 20x are described as five and twenty times the Pro per-session allowance, with no absolute number. The pool is shared across claude.ai and Claude Code, and "Usage credits don't affect this reset timing." Settings > Usage shows how much of the session you have used and the time remaining; Anthropic's pages do not say what starts the clock. The Codex window has been switched off and on. On 2026-07-12 Tibo Sottiaux announced "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans". In a post timestamped 2026-08-25 01:16 UTC he wrote "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex", which puts the return for Plus on 2026-08-25 US Pacific time; the same post keeps the limit "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. On the board, this window is not what is tracked. LAST RESET and NEXT EST. describe account-wide resets announced by the vendor, and the STATUS words RESET NOW, BANKED, TEASE, WAITING and OVERDUE never refer to your own window. To count down your own window, paste the time from /status or /usage into the window tracker; the math stays in your browser. What the official docs say - "A new window starts when you send your first message in Work or Codex after the previous window ends." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Your session-based usage limit will reset every five hours." — What is the Max plan? (Claude Help Center) (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans" — Tibo Sottiaux on X, 2026-07-12: 5-hour limit removed for Plus, Business and Pro (https://x.com/thsottiaux/status/2076365965915467978) - "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex." — Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit returns for Plus "tomorrow"; not enabled for Pro (https://x.com/thsottiaux/status/2092058556707344708) - "Usage credits don't affect this reset timing." — Manage usage credits for paid Claude plans (Claude Help Center) (https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans) FAQ: Q: Does the window start at a fixed hour of the day? A: Codex: no. "A new window starts when you send your first message in Work or Codex after the previous window ends." Claude: the help center says the session limit "will reset every five hours" and Settings > Usage shows the time remaining; Anthropic's pages do not state what starts the clock. Q: Does Codex Pro still have a 5-hour limit? A: Not as of 2026-08-25. Tibo Sottiaux: "We are for the upcoming months keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions." Plus got the limit back; the weekly limit still applies to Pro. Q: Does Grok have a 5-hour window? A: xAI's FAQ describes a weekly pool: "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." It does not mention a five-hour window, and this site does not add one. Q: If my 5-hour window clears, does my weekly limit refill? A: No. Codex: "Where both apply, you need allowance remaining in both to continue". Claude: the weekly limit "applies across all models" and resets on its own fixed weekly time. Clearing the short window does not touch the long one. Sources: 1. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 3. What is the Max plan? (Claude Help Center) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 4. Manage usage credits for paid Claude plans (Claude Help Center) — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 5. Usage limit best practices: Settings > Usage (Claude Help Center) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 6. Tibo Sottiaux on X, 2026-07-12: 5-hour limit removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit returns for Plus "tomorrow"; not enabled for Pro — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) 8. xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) — https://docs.x.ai/grok/faq (verified 2026-09-17) 9. Use Claude Code with your Pro or Max plan (Claude Help Center) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) ### 什么是 5 小时窗口？ URL: https://didreset.com/zh/guides/5-hour-window Language: zh · Kind: 术语 · Term code: five-hour-window · Published: 2026-09-18 · Updated: 2026-09-18 5 小时窗口是 Codex 与 Claude Code 按账号计量的短期额度。Codex：上一个窗口结束后你发出第一条消息时新窗口开始；Claude：会话限额每五小时刷新。它在独立的周限额之下，按你自己的时钟刷新，用 /status（Codex）或 /usage（Claude Code）查看。xAI 的 Grok FAQ 只描述了周用量池。 定义 Codex："Depending on your plan, usage limits may apply over a five-hour window and a weekly window. Where both apply, you need allowance remaining in both to continue"。窗口不是固定时钟：它从上一个窗口结束后你的第一条消息开始，而且 "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining."。定价页公布了各计划的估计值，例如 Plus 每个五小时周期 5–45 条 GPT-6 Astra 本地消息，但它们 "are not fixed message limits"。 Claude Pro 与 Max："Your session-based usage limit will reset every five hours."。Max 5x 与 20x 的描述是 Pro 每会话额度的 5 倍和 20 倍，没有绝对数字。池子在 claude.ai 与 Claude Code 之间共享，并且 "Usage credits don't affect this reset timing."。Settings > Usage 显示会话已用多少和剩余时间；Anthropic 的页面没有说明时钟从什么开始计。 Codex 的这个窗口关过又开过。2026-07-12 Tibo Sottiaux 宣布 "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans"。在时间戳为 2026-08-25 01:16 UTC 的帖子里他写道 "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex"，据此 Plus 的恢复日期是美国太平洋时间 2026-08-25；同一帖子说接下来几个月 Pro $100 与 Pro $200 保持 "not enabled"。 在出发板上，这个窗口不是被追踪的对象。「上次重置」和「预计下次」描述的是厂商公告的全员重置，状态词「已重置」「重置卡」「预告中」「等待中」「已超期」从不指你自己的窗口。要给自己的窗口倒计时，把 /status 或 /usage 里的时间贴进窗口倒计时工具；计算都在你的浏览器里完成。 官方文档怎么说 - "A new window starts when you send your first message in Work or Codex after the previous window ends." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "Your session-based usage limit will reset every five hours." — What is the Max plan? (Claude Help Center) (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans" — Tibo Sottiaux on X, 2026-07-12: 5-hour limit removed for Plus, Business and Pro (https://x.com/thsottiaux/status/2076365965915467978) - "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex." — Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit returns for Plus "tomorrow"; not enabled for Pro (https://x.com/thsottiaux/status/2092058556707344708) - "Usage credits don't affect this reset timing." — Manage usage credits for paid Claude plans (Claude Help Center) (https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans) FAQ: Q: 窗口是从一天里的固定钟点开始的吗？ A: Codex：不是。"A new window starts when you send your first message in Work or Codex after the previous window ends."。Claude：帮助中心说会话限额 "will reset every five hours"，Settings > Usage 显示剩余时间；Anthropic 的页面没有说明时钟从什么开始计。 Q: Codex Pro 还有 5 小时限制吗？ A: 截至 2026-08-25 没有。Tibo Sottiaux："We are for the upcoming months keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions."。Plus 恢复了这个限制；周限额对 Pro 仍然适用。 Q: Grok 有 5 小时窗口吗？ A: xAI 的 FAQ 描述的是周用量池："The usage pool limit resets every week on a schedule shown in the Usage tab in Settings."。它没有提到五小时窗口，本站也不替它加一个。 Q: 5 小时窗口清空后，周限额会跟着回来吗？ A: 不会。Codex："Where both apply, you need allowance remaining in both to continue"。Claude：周限额 "applies across all models"，按自己固定的每周时间刷新。短窗口清空不会动长窗口。 Sources: 1. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 2. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 3. What is the Max plan? (Claude Help Center) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 4. Manage usage credits for paid Claude plans (Claude Help Center) — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 5. Usage limit best practices: Settings > Usage (Claude Help Center) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 6. Tibo Sottiaux on X, 2026-07-12: 5-hour limit removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit returns for Plus "tomorrow"; not enabled for Pro — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) 8. xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) — https://docs.x.ai/grok/faq (verified 2026-09-17) 9. Use Claude Code with your Pro or Max plan (Claude Help Center) — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) ### What is a weekly limit? URL: https://didreset.com/guides/weekly-limit Language: en · Kind: Term · Term code: weekly-limit · Published: 2026-09-18 · Updated: 2026-09-18 A weekly limit is the second, longer cap on subscription usage. Codex: "Weekly limits may also apply", with no published number. Claude Pro and Max: "The weekly limit resets at a fixed time each week that is assigned to your account." Grok: "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." None of the three vendors publishes a weekly figure; your own reset time is in /status, /usage or Settings → Usage. Definition Codex: the weekly limit sits on top of the 5-hour window and "controls how much included work you can use over the weekly usage period." The clock is per account: "Your new weekly period starts with your first Work or Codex request after the reset, and your next automatic weekly reset is seven days after that request." A full banked reset "changes your weekly reset date". /status prints the date, for example "Weekly limit: … 97% left (resets 17:41 on 1 Jul)". Claude: Anthropic announced weekly limits on 2025-07-28: "We're rolling out new weekly rate limits for Claude Pro and Max in late August. We estimate they'll apply to less than 5% of subscribers". The limit "applies across all models" and "resets at a fixed time each week that is assigned to your account"; "Your reset day and time stay the same regardless of when you start using Claude or when your subscription begins". Settings > Usage shows separate weekly bars "for Opus only and all other models." Grok: paid usage is metered from a weekly pool. "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings", and "When your weekly usage limit is met, paid features will pause until your weekly limit resets." Grok Bot has its own weekly usage: "Usage resets weekly." None of the three vendors publishes a weekly number, and this site does not estimate one. On the board, an account-wide reset usually means weekly limits were refreshed for everyone: @bot's post of 2026-08-26 says "We're also resetting weekly usage limits for all users", and a Codex global reset refreshes the limits named in its announcement. Your own weekly clock stays separate: the board never shows it, and RESET NOW, BANKED, TEASE, WAITING and OVERDUE describe announcements, not your account. What the official docs say - "Weekly limits may also apply." — Codex pricing and usage limits (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "The weekly limit resets at a fixed time each week that is assigned to your account." — What is the Max plan? (Claude Help Center) (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "your next automatic weekly reset is seven days after that request" — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) - "We're rolling out new weekly rate limits for Claude Pro and Max in late August." — Anthropic on X, 2025-07-28: weekly rate limits for Pro and Max (https://x.com/AnthropicAI/status/1949898502688903593) FAQ: Q: Does clearing the 5-hour window give back weekly usage? A: No. Codex: "Where both apply, you need allowance remaining in both to continue". Claude: the weekly limit "applies across all models" on its own fixed weekly time. The short window and the long window are separate meters. Q: When exactly does my weekly limit reset? A: Codex: the "Weekly limit" line in /status, for example "(resets 17:41 on 1 Jul)". Claude: /usage in Claude Code or Settings > Usage on claude.ai, which shows the reset "for Opus only and all other models." Grok: the schedule in the Usage tab in Settings. There is no universal time. Q: How much usage is in a week? A: Not published. Codex's pricing page says only "Weekly limits may also apply"; Anthropic gives no count; xAI's FAQ gives a schedule but no amount. This site marks those cells "Not published" instead of guessing. Q: Does an official reset refill my weekly limit? A: It depends on the announcement. A Codex full banked reset "refreshes your 5-hour and weekly Codex usage windows"; a global reset "refreshes the eligible usage limits specified in the announcement"; @bot on 2026-08-26: "We're also resetting weekly usage limits for all users." Read the scope on the board row and its linked post. Sources: 1. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. What is the Max plan? (Claude Help Center) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 4. What is the Pro plan? (Claude Help Center) — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 5. Anthropic on X, 2025-07-28: weekly rate limits for Pro and Max — https://x.com/AnthropicAI/status/1949898502688903593 (verified 2026-09-17) 6. Usage limit best practices: Settings > Usage (Claude Help Center) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 7. xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) — https://docs.x.ai/grok/faq (verified 2026-09-17) 8. xAI docs: Grok Bot overview ("Usage resets weekly.") — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 9. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 10. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 11. openai/codex issue #29948: /status output with 5h limit and Weekly limit bars — https://github.com/openai/codex/issues/29948 (verified 2026-09-17) ### 什么是周限额？ URL: https://didreset.com/zh/guides/weekly-limit Language: zh · Kind: 术语 · Term code: weekly-limit · Published: 2026-09-18 · Updated: 2026-09-18 周限额是第二道、更长周期的用量上限。Codex 定价页只说周限额「可能也适用」，没有数字；Claude Pro 与 Max 的周限额在分配给你账号的每周固定时间刷新；Grok 的周用量池按 Settings 里 Usage 标签页的时间表刷新。三家都不公布周额度数字；自己的刷新时间看 /status、/usage 或 Settings。 定义 Codex：周限额叠在 5 小时窗口之上，"controls how much included work you can use over the weekly usage period."。时钟按账号计："Your new weekly period starts with your first Work or Codex request after the reset, and your next automatic weekly reset is seven days after that request."。完整重置卡 "changes your weekly reset date"。/status 会打印日期，例如 "Weekly limit: … 97% left (resets 17:41 on 1 Jul)"。 Claude：Anthropic 在 2025-07-28 宣布周限额："We're rolling out new weekly rate limits for Claude Pro and Max in late August. We estimate they'll apply to less than 5% of subscribers"。这个限额 "applies across all models"，并且 "resets at a fixed time each week that is assigned to your account"；"Your reset day and time stay the same regardless of when you start using Claude or when your subscription begins"。Settings > Usage 分别显示 "for Opus only and all other models" 两条周用量条。 Grok：付费用量从一个周用量池计量。"The usage pool limit resets every week on a schedule shown in the Usage tab in Settings"，并且 "When your weekly usage limit is met, paid features will pause until your weekly limit resets."。Grok Bot 有自己的周用量："Usage resets weekly."。三家厂商都不公布周额度数字，本站也不估算。 在出发板上，全员重置通常意味着所有人的周限额被刷新了：@bot 2026-08-26 的帖子说 "We're also resetting weekly usage limits for all users"，Codex 的全局重置刷新公告里点名的额度。你自己的周时钟仍然是独立的：出发板从不显示它，「已重置」「重置卡」「预告中」「等待中」「已超期」描述的是公告，不是你的账号。 官方文档怎么说 - "Weekly limits may also apply." — Codex pricing and usage limits (learn.chatgpt.com) (https://learn.chatgpt.com/docs/pricing) - "The weekly limit resets at a fixed time each week that is assigned to your account." — What is the Max plan? (Claude Help Center) (https://support.claude.com/en/articles/11049741-what-is-the-max-plan) - "your next automatic weekly reset is seven days after that request" — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings." — xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) (https://docs.x.ai/grok/faq) - "We're rolling out new weekly rate limits for Claude Pro and Max in late August." — Anthropic on X, 2025-07-28: weekly rate limits for Pro and Max (https://x.com/AnthropicAI/status/1949898502688903593) FAQ: Q: 5 小时窗口清空后，周额度会回来吗？ A: 不会。Codex："Where both apply, you need allowance remaining in both to continue"。Claude：周限额 "applies across all models"，按自己固定的每周时间刷新。短窗口和长窗口是两个独立的计量器。 Q: 我的周限额到底几点刷新？ A: Codex：/status 里的 "Weekly limit" 行，例如 "(resets 17:41 on 1 Jul)"。Claude：Claude Code 的 /usage 或 claude.ai 的 Settings > Usage，后者分别显示 "for Opus only and all other models" 的刷新时间。Grok：Settings 里 Usage 标签页的时间表。没有全站统一的时间。 Q: 一周的额度有多少？ A: 未公布。Codex 定价页只说 "Weekly limits may also apply"；Anthropic 不给条数；xAI 的 FAQ 给了时间表但没有数量。本站把这些格子标为「未公布」，不猜。 Q: 官方重置会把我的周限额补回来吗？ A: 看公告。Codex 完整重置卡 "refreshes your 5-hour and weekly Codex usage windows"；全局重置 "refreshes the eligible usage limits specified in the announcement"；@bot 2026-08-26："We're also resetting weekly usage limits for all users."。以出发板那一行的范围和它链接的原帖为准。 Sources: 1. Codex pricing and usage limits (learn.chatgpt.com) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. What is the Max plan? (Claude Help Center) — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 4. What is the Pro plan? (Claude Help Center) — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 5. Anthropic on X, 2025-07-28: weekly rate limits for Pro and Max — https://x.com/AnthropicAI/status/1949898502688903593 (verified 2026-09-17) 6. Usage limit best practices: Settings > Usage (Claude Help Center) — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 7. xAI docs: Grok website and apps FAQ (Settings → Usage, weekly pool) — https://docs.x.ai/grok/faq (verified 2026-09-17) 8. xAI docs: Grok Bot overview ("Usage resets weekly.") — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 9. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 10. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 11. openai/codex issue #29948: /status output with 5h limit and Weekly limit bars — https://github.com/openai/codex/issues/29948 (verified 2026-09-17) ### What is a signal? URL: https://didreset.com/guides/signal Language: en · Kind: Term · Term code: signal · Published: 2026-09-18 · Updated: 2026-09-18 A signal is an official post about usage limits from an account this board watches, filed as one of five kinds: reset, banked, boost, promo or tease. Only a reset moves LAST RESET; a banked reset or a tease changes the STATUS word for 72 hours. Every signal links to its original post, for example @bot's 2026-08-26 "We're also resetting weekly usage limits for all users." Alerts fire on signals, and teases are marked unconfirmed. Definition A signal is the unit this site records. It is an official statement, usually an X post from a vendor account or a help-center change, that says something about usage limits for many accounts. Each signal is stored with the announcing account, the announcement time in UTC, its scope (everyone or a plan), a short excerpt and the link to the original post. Rows not yet checked against the original carry the label "unverified". There are five kinds. reset: an account-wide refresh, such as the Codex global reset OpenAI documents for 2026-09-07, applied immediately and not banked. banked: a card drop you redeem yourself, such as the one Tibo Sottiaux described on 2026-07-13: "You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you." boost: a limit raised. promo: a temporary promotion. tease: a reset hinted at but not yet in effect. What is not a signal: your own 5-hour window or weekly limit refreshing, an error message on your screen, or a rule change without a reset. Tibo Sottiaux's 2026-08-25 post about bringing back the 5h limit for Plus is a rule change; it belongs on the limits page, not on the board. OpenAI's help center draws the same line for its own resets: "An automatic or global reset refreshes the eligible usage limits specified in the announcement." On the board, the STATUS flap is derived from the latest signals: RESET NOW when a reset was announced in the last 24 hours; BANKED or TEASE when the latest signal is a banked drop or a tease from the last 72 hours; WAITING when nothing new has happened since the last reset; OVERDUE when the wait has passed 1.5 times the median gap of the last 10 resets; NO DATA when nothing is recorded. Alerts: email and browser push go out for reset, banked and tease signals, with teases marked unconfirmed; SMS goes out for reset and banked, and a phone call only for a confirmed reset. What the official docs say - "We're also resetting weekly usage limits for all users." — Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users (https://x.com/bot/status/2092664183334846840) - "An automatic or global reset refreshes the eligible usage limits specified in the announcement." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you." — Tibo Sottiaux on X, 2026-07-13: applying a banked reset replenishes weekly usage (https://x.com/thsottiaux/status/2076735790567338203) FAQ: Q: Is a tease a reset? A: No. A tease is an official hint that a reset is coming, not yet in effect. The board shows TEASE for 72 hours, LAST RESET does not move, and the free email and push alerts for it carry an "unconfirmed" label. SMS and phone calls are reserved for confirmed signals. Q: Why does the board say WAITING when I was just rate-limited? A: Because your own window is not a signal. WAITING means no vendor has announced anything since the last account-wide reset. Your 5-hour window and weekly limit refresh on your own clock; check them with /status, /usage or Settings → Usage, and use the window tracker for a countdown. Q: What does "unverified" mean on a row? A: The row was ingested from a watched account but has not yet been checked by hand against the original post. It still links to the source so you can read it yourself; the flag clears once the excerpt, time and scope have been confirmed. Q: Where do signals come from? A: From the official accounts this board watches: @thsottiaux for Codex; @ClaudeDevs and @lydiahallie for Claude; @grok, @bot and @elonmusk for Grok; and from vendor help pages when they document a reset, such as OpenAI's entry for the 2026-09-07 global reset. The public status API and RSS feed expose the same records. Sources: 1. DidReset public status API: board status and cadence fields — https://didreset.com/api/status (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-07-13: applying a banked reset replenishes weekly usage — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) 5. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 6. DidReset fares: which alert channels fire on which signals — https://didreset.com/pricing (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit returns for Plus "tomorrow"; not enabled for Pro — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ### 什么是信号？ URL: https://didreset.com/zh/guides/signal Language: zh · Kind: 术语 · Term code: signal · Published: 2026-09-18 · Updated: 2026-09-18 信号是本站盯着的官方账号发布的、关于额度限制的帖子，归入五类之一：重置、重置卡、提额、促销、预告。只有重置会移动「上次重置」；重置卡发放或预告会在 72 小时内改变状态词。每条信号都链接到原帖，例如 @bot 2026-08-26 宣布为所有用户重置周限额。提醒按信号触发，预告标明「未确认」。 定义 信号是本站记录的基本单位。它是一条官方陈述，通常是厂商账号在 X 上的帖子或帮助中心的变更，内容涉及一大批账号的额度限制。每条信号都记录公告账号、UTC 公告时间、范围（全员或某个计划）、一段摘录，以及原帖链接。尚未与原文核对的行带有「待核实」标签。 五类。重置（reset）：全员刷新，例如 OpenAI 帮助中心记载的 2026-09-07 Codex 全局重置，立即生效、不存为重置卡。重置卡（banked）：需要你自己兑换的发卡，例如 Tibo Sottiaux 2026-07-13 描述的那次："You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you."。提额（boost）：上限被调高。促销（promo）：临时活动。预告（tease）：官方暗示会有重置，但尚未生效。 不算信号的：你自己的 5 小时窗口或周限额刷新、你屏幕上的报错，以及没有重置的规则变更。Tibo Sottiaux 2026-08-25 关于 Plus 恢复 5 小时限制的帖子是规则变更，归额度上限页，不上出发板。OpenAI 帮助中心对自家的重置也划了同一条线："An automatic or global reset refreshes the eligible usage limits specified in the announcement." 在出发板上，状态翻牌由最新的信号推导：24 小时内公告了重置显示「已重置」（RESET NOW）；最近一条信号是 72 小时内的发卡或预告，显示「重置卡」（BANKED）或「预告中」（TEASE）；上次重置以来没有新动静，显示「等待中」（WAITING）；等待时间超过近 10 次重置中位间隔的 1.5 倍，显示「已超期」（OVERDUE）；没有任何记录显示「无数据」（NO DATA）。提醒：重置、重置卡、预告三类信号都发邮件和浏览器推送，预告标明未确认；短信只发重置和重置卡，电话只在确认的重置时打。 官方文档怎么说 - "We're also resetting weekly usage limits for all users." — Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users (https://x.com/bot/status/2092664183334846840) - "An automatic or global reset refreshes the eligible usage limits specified in the announcement." — Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) (https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex) - "You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you." — Tibo Sottiaux on X, 2026-07-13: applying a banked reset replenishes weekly usage (https://x.com/thsottiaux/status/2076735790567338203) FAQ: Q: 预告算重置吗？ A: 不算。预告是官方暗示重置将至，但尚未生效。出发板显示「预告中」72 小时，「上次重置」不动，为它发出的免费邮件和推送带「未确认」标签。短信和电话只留给确认的信号。 Q: 我刚被限流，为什么出发板显示「等待中」？ A: 因为你自己的窗口不是信号。「等待中」表示自上次全员重置以来没有任何厂商发过公告。你的 5 小时窗口和周限额按自己的时钟刷新；用 /status、/usage 或 Settings → Usage 查看，用窗口倒计时工具倒数。 Q: 某一行上的「待核实」是什么意思？ A: 这一行是从被监视的账号抓进来的，但还没有人工对照原帖核对。它仍然链接到来源，你可以自己去读；摘录、时间和范围确认后标签会消失。 Q: 信号从哪里来？ A: 来自本站盯着的官方账号：Codex 的 @thsottiaux；Claude 的 @ClaudeDevs 与 @lydiahallie；Grok 的 @grok、@bot 与 @elonmusk；以及厂商帮助页记载的重置，例如 OpenAI 关于 2026-09-07 全局重置的条目。公开状态 API 和 RSS 订阅暴露同样的记录。 Sources: 1. DidReset public status API: board status and cadence fields — https://didreset.com/api/status (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex (OpenAI Help Center, revision 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 4. Tibo Sottiaux on X, 2026-07-13: applying a banked reset replenishes weekly usage — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) 5. How banked Codex resets work (OpenAI Help Center) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 6. DidReset fares: which alert channels fire on which signals — https://didreset.com/pricing (verified 2026-09-17) 7. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit returns for Plus "tomorrow"; not enabled for Pro — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) ## Usage limits by plan ### Codex usage limits URL: https://didreset.com/limits/codex Language: en · Published: 2026-09-18 · Updated: 2026-09-18 OpenAI publishes only five-hour estimates for Codex: on Plus, 5–45 GPT-6 Astra local messages per five-hour period, with Pro 5x, Pro 20x and Business columns on the same table. No plan has a published weekly number; the pricing page says weekly limits "may also apply". Free and Go have no per-window estimate. The last account-wide Codex reset on this board was announced {lastReset}. What are the limits on each plan? - Free · $0/month · 5-hour window: Not published (varies-by-plan) · Source: https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan · Verified: 2026-09-17 - Free · $0/month · Weekly limit: Not published · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Go · $8/month · 5-hour window: Not published (varies-by-plan) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Go · $8/month · Weekly limit: Not published · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Plus · $20/month · 5-hour window: GPT-6 Astra 5–45 · GPT-5.6 Sol 10–100 · GPT-5.6 Terra 25–200 · GPT-5.6 Luna 250–2,000 (estimate) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Plus · $20/month · Weekly limit: Not published (weekly-may-apply) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Pro 5x · $100/month · 5-hour window: GPT-6 Astra 25–225 · GPT-5.6 Sol 50–500 · GPT-5.6 Terra 125–1,000 · GPT-5.6 Luna 1,250–10,000 (pro-5h-off) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Pro 5x · $100/month · Weekly limit: Not published (weekly-may-apply) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Pro 20x · $200/month · 5-hour window: GPT-6 Astra 100–900 · GPT-5.6 Sol 200–2,000 · GPT-5.6 Terra 500–4,000 · GPT-5.6 Luna 5,000–40,000 (pro-200-price) · Source: https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex · Verified: 2026-09-17 - Pro 20x · $200/month · Weekly limit: Not published (weekly-may-apply) · Source: https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex · Verified: 2026-09-17 - Business Standard seat · $20/user/month (annual, 2+ users) · 5-hour window: GPT-6 Astra 5–45 · GPT-5.6 Sol 10–100 · GPT-5.6 Terra 25–200 · GPT-5.6 Luna 250–2,000 (business-same-as-plus) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Business Standard seat · $20/user/month (annual, 2+ users) · Weekly limit: Not published (weekly-may-apply) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Enterprise / Edu · contact sales · 5-hour window: Not published (credits) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 - Enterprise / Edu · contact sales · Weekly limit: Not published (credits) · Source: https://learn.chatgpt.com/docs/pricing · Verified: 2026-09-17 Cells the vendor has not published are marked "Not published". This site does not estimate them. What has changed? - 2026-09-16: Help-center revision documents purchased instant resets for eligible personal Plus and Pro accounts: refreshes the five-hour and weekly allowance at checkout, "cannot be saved for later", and "brings your weekly allowance forward". — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex - 2026-09-09: Some banked resets failed to fully apply in ChatGPT Work and Codex during a time window; affected users received a replacement reset and an apology email. — https://x.com/thsottiaux/status/2097752790177370535 - 2026-09-07: Global automatic reset applied to Plus, Pro and Business usage limits; applied immediately, not saved as a banked reset. — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex - 2026-09-04: GPT-6 Astra rolled out to Plus, Pro and Business in Work and Codex ahead of schedule, drawing from the shared allowance; full banked reset for Plus, Pro and Business the same day (post timestamped 2026-09-05 00:39 UTC, Sep 4 PT). — https://x.com/thsottiaux/status/2096035437299237298 - 2026-09-03: Astra delay compensation: one banked reset per day without Astra access on a paid ChatGPT plan; drops on Sep 3 (cutoff 10 PM PT) and Sep 4 (cutoff 8 PM PT) for Plus, Pro and Business, including Standard and Premium seats. — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work - 2026-08-25: Tibo Sottiaux: "Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex." The post is timestamped 2026-08-25 01:16 UTC (Aug 24 evening PT), so the return day is recorded here by the post date; no source states an exact calendar day. Pro $100 and Pro $200 keep the 5h limit "not enabled" for the upcoming months. — https://x.com/thsottiaux/status/2092058556707344708 - 2026-08-21: 20M active users: a banked reset credited to every paid Codex and ChatGPT Work user (landed by 8pm PT; confirmed 2026-08-22). — https://x.com/thsottiaux/status/2090766694897619318 - 2026-07-13: First banked reset granted to every ChatGPT Work and Codex account at the 7M active-users milestone; applying it "will replenish the weekly usage for you". — https://x.com/thsottiaux/status/2076735790567338203 - 2026-07-12: "Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans" (weekly limit still applies); banked-reset redemption extended from the desktop app to web and mobile. — https://x.com/thsottiaux/status/2076365965915467978 - 2026-06-18: First "sneaky double reset": a full reset for everyone plus one reset deposited into the reset bank to use at leisure. — https://x.com/thsottiaux/status/2067399435009622521 - 2026-06-11: Changelog: rate-limit reset banking added for Plus and Pro users, "including one free reset at launch and referral invitations for earning more". — https://learn.chatgpt.com/docs/changelog - 2026-04-09: $100 Pro plan launched "with a ton of codex usage"; limits reset alongside. — https://x.com/thsottiaux/status/2042299371602264319 - 2025-12-25: Holiday promotion: rate limits reset and usage limits lifted to 2X the usual until 2026-01-01. — https://x.com/thsottiaux/status/2004100061933064395 - 2025-11-07: Codex usage limits increased by 50% for Plus, Edu and Team; gpt-5-codex-mini released with "up to 4X more usage included"; Pro gets priority processing. — https://x.com/thsottiaux/status/1986863197803192782 How the windows work Codex meters a subscription in two windows. OpenAI’s help center: "Depending on your plan, usage limits may apply over a five-hour window and a weekly window." Where both apply you need allowance remaining in both, so clearing the five-hour window does not refill the weekly one. Local CLI and IDE messages, cloud chats and ChatGPT Work draw from one shared allowance, and "Switching models does not restore allowance in a shared usage pool." The five-hour limit has not been constant. Tibo Sottiaux removed it for Plus, Business and Pro on 2026-07-12, then wrote on 2026-08-25 (01:16 UTC) that it would return for Plus accounts "tomorrow" while staying "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. The weekly window runs on a per-account clock: after a purchased reset, "your next automatic weekly reset is seven days after that request". The estimates in the table "are not fixed message limits"; OpenAI says to "check your usage dashboard for current limits and reset times". Inside the CLI, /status shows both windows and their reset times, and /usage opens the usage menu where an earned banked reset can be redeemed. "Using a full banked reset refreshes your 5-hour and weekly Codex usage windows and changes your weekly reset date." FAQ: Q: Does the Codex 5-hour limit apply to Pro plans right now? A: Not according to the latest official statement on record. On 2026-07-12 Tibo Sottiaux temporarily removed the 5-hour limit for Plus, Business and Pro; on 2026-08-25 he wrote it would return for Plus the next day while staying "not enabled for Pro $100 and Pro $200 subscriptions" for the upcoming months. The Pro rows in the table are OpenAI’s published per-five-hour estimates, and the weekly limit still applies. Q: What is the Codex weekly limit? A: OpenAI has not published a number for any plan. The pricing page says "Weekly limits may also apply", and the help center explains the clock: your new weekly period starts with your first Work or Codex request after a reset, and the next automatic weekly reset is seven days after that request. Your own reset time is shown in Settings → Usage, on the usage dashboard, or by /status in the CLI. Q: Do banked resets or purchased resets change these limits? A: No. A banked reset is "a one-time Codex usage-limit reset saved to your account until you use it or it expires"; it "does not permanently increase your plan's usage limit". A purchased instant reset refreshes the five-hour and weekly allowance as soon as checkout succeeds and "cannot be saved for later". Neither raises the numbers in the table; they only refill the windows once. Q: Where do these numbers come from? A: Every row links to the OpenAI page it was read from, with the date it was verified. Five-hour estimates come from the Codex pricing page and the help-center article on GPT-6 Astra; plan changes come from Tibo Sottiaux’s posts, the ChatGPT changelog and the help center. Cells OpenAI has not published are marked "Not published"; this site does not estimate them. Sources: 1. Pricing | ChatGPT Learn (Codex) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex | OpenAI Help Center (read via Wayback snapshot dated 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Using Codex with your ChatGPT plan | OpenAI Help Center (read via Wayback snapshot dated 2026-09-15) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 4. How banked Codex resets work | OpenAI Help Center (read via Wayback snapshot dated 2026-09-09) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 5. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit back for Plus "tomorrow", stays off for Pro — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) 6. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 7. Tibo Sottiaux on X, 2025-11-07: usage limits +50% for Plus, Edu and Team — https://x.com/thsottiaux/status/1986863197803192782 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2025-12-25: holiday reset and 2X limits until 2026-01-01 — https://x.com/thsottiaux/status/2004100061933064395 (verified 2026-09-17) 9. Tibo Sottiaux on X, 2026-04-09: $100 Pro plan launched — https://x.com/thsottiaux/status/2042299371602264319 (verified 2026-09-17) 10. ChatGPT & Codex changelog | ChatGPT Learn (2026-06-11: reset banking) — https://learn.chatgpt.com/docs/changelog (verified 2026-09-17) 11. Tibo Sottiaux on X, 2026-06-18: "sneaky double reset" — https://x.com/thsottiaux/status/2067399435009622521 (verified 2026-09-17) 12. Tibo Sottiaux on X, 2026-07-13: first banked reset for every Work and Codex account — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) 13. Tibo Sottiaux on X, 2026-08-21: 20M active users, banked reset for every paid user — https://x.com/thsottiaux/status/2090766694897619318 (verified 2026-09-17) 14. Tibo Sottiaux on X, 2026-09-05 00:39 UTC (Sep 4 PT): GPT-6 Astra rollout and full banked reset — https://x.com/thsottiaux/status/2096035437299237298 (verified 2026-09-17) 15. Tibo Sottiaux on X, 2026-09-09: some banked resets failed to apply, replacement issued — https://x.com/thsottiaux/status/2097752790177370535 (verified 2026-09-17) 16. Developer commands | ChatGPT Learn (Codex CLI): /status and /usage — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) ### Codex 额度上限 URL: https://didreset.com/zh/limits/codex Language: zh · Published: 2026-09-18 · Updated: 2026-09-18 OpenAI 只公布 Codex 的 5 小时估算：Plus 每 5 小时约 5–45 条 GPT-6 Astra 本地消息，Pro 5x、20x、Business 同表列出；周限额各计划均未公布数字，定价页只写「可能同样适用」；Free、Go 无估算。板上 Codex 上次全员重置公告于 {lastReset}。 各计划的额度是多少？ - Free · $0/month · 5 小时窗口: 未公布 (varies-by-plan) · 来源: https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan · 核实: 2026-09-17 - Free · $0/month · 周限额: 未公布 · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Go · $8/month · 5 小时窗口: 未公布 (varies-by-plan) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Go · $8/month · 周限额: 未公布 · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Plus · $20/month · 5 小时窗口: GPT-6 Astra 5–45 · GPT-5.6 Sol 10–100 · GPT-5.6 Terra 25–200 · GPT-5.6 Luna 250–2,000 (estimate) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Plus · $20/month · 周限额: 未公布 (weekly-may-apply) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Pro 5x · $100/month · 5 小时窗口: GPT-6 Astra 25–225 · GPT-5.6 Sol 50–500 · GPT-5.6 Terra 125–1,000 · GPT-5.6 Luna 1,250–10,000 (pro-5h-off) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Pro 5x · $100/month · 周限额: 未公布 (weekly-may-apply) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Pro 20x · $200/month · 5 小时窗口: GPT-6 Astra 100–900 · GPT-5.6 Sol 200–2,000 · GPT-5.6 Terra 500–4,000 · GPT-5.6 Luna 5,000–40,000 (pro-200-price) · 来源: https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex · 核实: 2026-09-17 - Pro 20x · $200/month · 周限额: 未公布 (weekly-may-apply) · 来源: https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex · 核实: 2026-09-17 - Business Standard seat · $20/user/month (annual, 2+ users) · 5 小时窗口: GPT-6 Astra 5–45 · GPT-5.6 Sol 10–100 · GPT-5.6 Terra 25–200 · GPT-5.6 Luna 250–2,000 (business-same-as-plus) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Business Standard seat · $20/user/month (annual, 2+ users) · 周限额: 未公布 (weekly-may-apply) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Enterprise / Edu · contact sales · 5 小时窗口: 未公布 (credits) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 - Enterprise / Edu · contact sales · 周限额: 未公布 (credits) · 来源: https://learn.chatgpt.com/docs/pricing · 核实: 2026-09-17 厂商未公布的单元格写「未公布」，本站不估算。 改过什么？ - 2026-09-16: 帮助中心修订版记录了面向符合条件的个人 Plus 与 Pro 账号的付费即时重置：付款成功即刷新 5 小时与周额度，「不能留到以后」，并「把周额度周期提前」。 — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex - 2026-09-09: 某时间段内部分重置卡在 ChatGPT Work 与 Codex 中未完全生效；受影响用户收到补发的重置与致歉邮件。 — https://x.com/thsottiaux/status/2097752790177370535 - 2026-09-07: 对 Plus、Pro、Business 的用量限额执行全局自动重置；立即生效，不存为重置卡。 — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex - 2026-09-04: GPT-6 Astra 提前向 Work 与 Codex 的 Plus、Pro、Business 推出，用量从共享额度扣除；同日为 Plus、Pro、Business 发放完整重置卡（原帖时间戳 2026-09-05 00:39 UTC，即太平洋时间 9 月 4 日）。 — https://x.com/thsottiaux/status/2096035437299237298 - 2026-09-03: Astra 延期补偿：付费 ChatGPT 计划每无法使用 Astra 一天补一张重置卡；9 月 3 日（截止太平洋时间晚 10 点）与 9 月 4 日（截止晚 8 点）发放给 Plus、Pro、Business（含 Standard 与 Premium 席位）。 — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work - 2026-08-25: Tibo Sottiaux：「明天我们会为 ChatGPT Work 与 Codex 的 Plus 账号恢复 5 小时限制。」原帖时间戳为 2026-08-25 01:16 UTC（太平洋时间 8 月 24 日晚），本表按发帖日期记录；没有来源写明确切日历日。Pro $100 与 Pro $200 在接下来几个月继续「不启用」5 小时限制。 — https://x.com/thsottiaux/status/2092058556707344708 - 2026-08-21: 活跃用户达 2000 万：每位付费 Codex 与 ChatGPT Work 用户获得一张重置卡（太平洋时间晚 8 点前到账；2026-08-22 确认）。 — https://x.com/thsottiaux/status/2090766694897619318 - 2026-07-13: 活跃用户达 700 万，每个 ChatGPT Work 与 Codex 账号获得第一张重置卡；使用后「会为你补满周用量」。 — https://x.com/thsottiaux/status/2076735790567338203 - 2026-07-12: 「暂时取消所有 Plus、Business、Pro 计划的 5 小时用量限制」（周限额仍然生效）；重置卡兑换从桌面 App 扩展到网页与移动端。 — https://x.com/thsottiaux/status/2076365965915467978 - 2026-06-18: 第一次「偷偷双重重置」：全员完整重置，另存一次重置进重置卡库，可随时使用。 — https://x.com/thsottiaux/status/2067399435009622521 - 2026-06-11: 更新日志：为 Plus 与 Pro 用户加入重置卡（banked reset）机制，「上线时赠送一次重置，并可通过推荐邀请获得更多」。 — https://learn.chatgpt.com/docs/changelog - 2026-04-09: $100 的 Pro 计划上线，「附带大量 Codex 用量」；同时重置限额。 — https://x.com/thsottiaux/status/2042299371602264319 - 2025-12-25: 节日活动：限额重置，额度提高到平时的 2 倍，持续到 2026-01-01。 — https://x.com/thsottiaux/status/2004100061933064395 - 2025-11-07: Codex 的 Plus、Edu、Team 计划额度提高 50%；发布 gpt-5-codex-mini，「包含最多 4 倍用量」；Pro 获得优先处理。 — https://x.com/thsottiaux/status/1986863197803192782 窗口怎么算 Codex 按两个窗口计量订阅用量。OpenAI 帮助中心写道：「视计划而定，用量限制可能同时按 5 小时窗口和周窗口生效。」两者都生效时，两个窗口都要有余量，所以 5 小时窗口刷新不会补回周限额。本地 CLI / IDE 消息、云端对话与 ChatGPT Work 共用一份额度，「在共享用量池里切换模型不会恢复额度」。 5 小时限制并非一直存在。Tibo Sottiaux 于 2026-07-12 为 Plus、Business、Pro 取消了它，又在 2026-08-25（01:16 UTC）写道「明天」会为 Plus 账号恢复，而 Pro $100 与 Pro $200 在接下来几个月继续「不启用」。周窗口按账号自己的时钟运行：付费重置之后，「下一次自动周重置在那次请求的七天之后」。 表里的估算「不是固定的消息上限」，OpenAI 让用户「到用量面板查看当前限额与重置时间」。在 CLI 里，/status 显示两个窗口及其重置时间，/usage 打开用量菜单，可以在那里兑换已获得的重置卡。「使用一张完整重置卡会刷新你的 5 小时与周 Codex 用量窗口，并改变你的周重置日期。」 FAQ: Q: Codex 的 5 小时限制现在对 Pro 计划生效吗？ A: 按有记录的最新官方说法，不生效。Tibo Sottiaux 于 2026-07-12 暂时取消了 Plus、Business、Pro 的 5 小时限制；2026-08-25 他写道次日会为 Plus 恢复，而 Pro $100 与 Pro $200 在接下来几个月继续「不启用」。表中 Pro 行是 OpenAI 公布的每 5 小时估算，周限额仍然生效。 Q: Codex 的周限额是多少？ A: OpenAI 没有公布任何计划的周限额数字。定价页只写「周限制可能同样适用」；帮助中心解释了时钟：重置之后，你的新周期从第一条 Work 或 Codex 请求开始，下一次自动周重置在那条请求的七天之后。你自己的重置时间在 Settings → Usage、用量面板或 CLI 的 /status 里能看到。 Q: 重置卡或付费重置会改变这些上限吗？ A: 不会。重置卡是「一次性的 Codex 用量限额重置，保存在账号里直到使用或过期」，它「不会永久提高计划的用量上限」。付费即时重置在付款成功后立即刷新 5 小时与周额度，且「不能留到以后」。两者都不会提高表中的数字，只是把窗口补满一次。 Q: 这些数字从哪里来？ A: 每一行都链接到读取它的 OpenAI 页面，并标注核实日期。5 小时估算来自 Codex 定价页与关于 GPT-6 Astra 的帮助中心文章；计划变更来自 Tibo Sottiaux 的帖子、ChatGPT 更新日志与帮助中心。OpenAI 没有公布的单元格标为「未公布」，本站不估算。 Sources: 1. Pricing | ChatGPT Learn (Codex) — https://learn.chatgpt.com/docs/pricing (verified 2026-09-17) 2. Managing usage with GPT-6 Astra in Work and Codex | OpenAI Help Center (read via Wayback snapshot dated 2026-09-16) — https://help.openai.com/en/articles/20001516-managing-usage-with-gpt-6-astra-in-work-and-codex (verified 2026-09-17) 3. Using Codex with your ChatGPT plan | OpenAI Help Center (read via Wayback snapshot dated 2026-09-15) — https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan (verified 2026-09-17) 4. How banked Codex resets work | OpenAI Help Center (read via Wayback snapshot dated 2026-09-09) — https://help.openai.com/en/articles/20001498-how-banked-codex-resets-work (verified 2026-09-17) 5. Tibo Sottiaux on X, 2026-08-25 01:16 UTC: 5h limit back for Plus "tomorrow", stays off for Pro — https://x.com/thsottiaux/status/2092058556707344708 (verified 2026-09-17) 6. Tibo Sottiaux on X, 2026-07-12: 5-hour limit temporarily removed for Plus, Business and Pro — https://x.com/thsottiaux/status/2076365965915467978 (verified 2026-09-17) 7. Tibo Sottiaux on X, 2025-11-07: usage limits +50% for Plus, Edu and Team — https://x.com/thsottiaux/status/1986863197803192782 (verified 2026-09-17) 8. Tibo Sottiaux on X, 2025-12-25: holiday reset and 2X limits until 2026-01-01 — https://x.com/thsottiaux/status/2004100061933064395 (verified 2026-09-17) 9. Tibo Sottiaux on X, 2026-04-09: $100 Pro plan launched — https://x.com/thsottiaux/status/2042299371602264319 (verified 2026-09-17) 10. ChatGPT & Codex changelog | ChatGPT Learn (2026-06-11: reset banking) — https://learn.chatgpt.com/docs/changelog (verified 2026-09-17) 11. Tibo Sottiaux on X, 2026-06-18: "sneaky double reset" — https://x.com/thsottiaux/status/2067399435009622521 (verified 2026-09-17) 12. Tibo Sottiaux on X, 2026-07-13: first banked reset for every Work and Codex account — https://x.com/thsottiaux/status/2076735790567338203 (verified 2026-09-17) 13. Tibo Sottiaux on X, 2026-08-21: 20M active users, banked reset for every paid user — https://x.com/thsottiaux/status/2090766694897619318 (verified 2026-09-17) 14. Tibo Sottiaux on X, 2026-09-05 00:39 UTC (Sep 4 PT): GPT-6 Astra rollout and full banked reset — https://x.com/thsottiaux/status/2096035437299237298 (verified 2026-09-17) 15. Tibo Sottiaux on X, 2026-09-09: some banked resets failed to apply, replacement issued — https://x.com/thsottiaux/status/2097752790177370535 (verified 2026-09-17) 16. Developer commands | ChatGPT Learn (Codex CLI): /status and /usage — https://learn.chatgpt.com/docs/developer-commands?surface=cli (verified 2026-09-17) ### Claude Code usage limits URL: https://didreset.com/limits/claude Language: en · Published: 2026-09-18 · Updated: 2026-09-18 Anthropic publishes no message counts for Claude Code: the number of messages "will vary based on message length" and the model or feature used. What is published is the shape: a session limit that "will reset every five hours", a weekly limit across all models at a fixed time assigned to your account, and Max 5x and 20x at five and twenty times the Pro per-session allowance. The last account-wide Claude reset on this board was announced {lastReset}. What are the limits on each plan? - Pro · $20/month · 5-hour window: Not published (varies-by-length) · Source: https://support.claude.com/en/articles/8325606-what-is-the-pro-plan · Verified: 2026-09-17 - Pro · $20/month · Weekly limit: Not published (fixed-weekly) · Source: https://support.claude.com/en/articles/8325606-what-is-the-pro-plan · Verified: 2026-09-17 - Max 5x · $100/month · 5-hour window: 5× Pro (relative-to-pro) · Source: https://support.claude.com/en/articles/11049741-what-is-the-max-plan · Verified: 2026-09-17 - Max 5x · $100/month · Weekly limit: Not published (fixed-weekly) · Source: https://support.claude.com/en/articles/11049741-what-is-the-max-plan · Verified: 2026-09-17 - Max 20x · $200/month · 5-hour window: 20× Pro (relative-to-pro) · Source: https://support.claude.com/en/articles/11049741-what-is-the-max-plan · Verified: 2026-09-17 - Max 20x · $200/month · Weekly limit: Not published (discretion) · Source: https://support.claude.com/en/articles/11049741-what-is-the-max-plan · Verified: 2026-09-17 - Team Standard seat · $25/seat/month ($20 billed annually) · 5-hour window: Not published (per-seat) · Source: https://claude.com/pricing · Verified: 2026-09-17 - Team Standard seat · $25/seat/month ($20 billed annually) · Weekly limit: Not published (per-seat-weekly) · Source: https://code.claude.com/docs/en/costs · Verified: 2026-09-17 - Team Premium seat · $125/seat/month ($100 billed annually) · 5-hour window: 5× Standard seat (relative-to-standard) · Source: https://claude.com/pricing · Verified: 2026-09-17 - Team Premium seat · $125/seat/month ($100 billed annually) · Weekly limit: Not published (per-seat-weekly) · Source: https://code.claude.com/docs/en/costs · Verified: 2026-09-17 Cells the vendor has not published are marked "Not published". This site does not estimate them. What has changed? - 2025-07-28: Anthropic: "We're rolling out new weekly rate limits for Claude Pro and Max in late August", on top of the existing 5-hour session limit; estimated to apply to "less than 5% of subscribers". — https://x.com/AnthropicAI/status/1949898502688903593 How the windows work Claude Code on a subscription draws from the same pool as claude.ai: "all activity in both tools counts against the same usage limits". The session limit "will reset every five hours". Settings > Usage on claude.ai shows how much of the five-hour session you have used "plus the amount of time remaining in the session", and /usage inside Claude Code shows the same plan usage bars. The weekly limit was announced on 2025-07-28 for late August 2025, estimated to apply to "less than 5% of subscribers". It "applies across all models" and "resets at a fixed time each week that is assigned to your account"; the usage page also lists a separate weekly reset "for Opus only". Usage credits, where enabled, "don't affect this reset timing" for the five-hour limit. Team seats work the same way per seat: each member’s usage "draws from a per-seat allowance that resets on a rolling five-hour window and a weekly window", shared with Claude chat and Cowork, and the seat tier (Standard or Premium) sets its size. Anthropic has not published a message count or token figure for any tier, and reserves "weekly and monthly caps" at its discretion. FAQ: Q: Is there a published number of messages per five hours on Claude Pro or Max? A: No. Anthropic’s Pro article says the number of messages "will vary based on message length" and the model or feature you use. Max 5x and Max 20x are defined only relative to Pro: "five times" and "20 times the Pro plan's per-session usage allowance". Settings > Usage on claude.ai and /usage in Claude Code show your own remaining share and the time left in the session. Q: When does the Claude weekly limit reset? A: At a fixed time each week that is assigned to your account; Anthropic does not publish the day. The usage page shows when your weekly limit resets "for Opus only and all other models". Reaching the five-hour session limit does not move the weekly clock, and usage credits do not change the five-hour reset timing either. Q: Does Claude Code share limits with claude.ai and the IDE extensions? A: Yes. "Both Pro and Max plans offer usage limits that are shared across Claude and Claude Code", and your usage of all product surfaces "counts towards the same usage limit". On Team plans the per-seat allowance is likewise shared with Claude chat and Cowork. Q: Has Anthropic ever reset the weekly limit for everyone? A: The board above records each account-wide reset announced on Anthropic accounts, each linked to its original post, and the cadence strip below counts them. Anthropic does not announce resets in advance, and this site does not promise when the next one will happen; the estimate on the board is the last reset plus the median gap. Sources: 1. What is the Pro plan? | Claude Help Center — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 2. What is the Max plan? | Claude Help Center — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 3. Pricing | Claude (claude.com/pricing) — https://claude.com/pricing (verified 2026-09-17) 4. Manage costs effectively – Claude Code Docs — https://code.claude.com/docs/en/costs (verified 2026-09-17) 5. Usage limit best practices | Claude Help Center — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 6. Use Claude Code with your Pro or Max plan | Claude Help Center — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 7. Anthropic on X, 2025-07-28: weekly rate limits for Pro and Max announced — https://x.com/AnthropicAI/status/1949898502688903593 (verified 2026-09-17) 8. Manage usage credits for paid Claude plans | Claude Help Center — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 9. How do usage and length limits work? | Claude Help Center — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) ### Claude Code 额度上限 URL: https://didreset.com/zh/limits/claude Language: zh · Published: 2026-09-18 · Updated: 2026-09-18 Anthropic 未公布 Claude Code 消息条数：能发多少「会随消息长度」及模型、功能而变。公布的是结构：会话限额「每五小时重置」，周限额跨全部模型、按账号固定时间重置，Max 5x、20x 为 Pro 每会话额度的 5 倍、20 倍。板上 Claude 上次全员重置公告于 {lastReset}。 各计划的额度是多少？ - Pro · $20/month · 5 小时窗口: 未公布 (varies-by-length) · 来源: https://support.claude.com/en/articles/8325606-what-is-the-pro-plan · 核实: 2026-09-17 - Pro · $20/month · 周限额: 未公布 (fixed-weekly) · 来源: https://support.claude.com/en/articles/8325606-what-is-the-pro-plan · 核实: 2026-09-17 - Max 5x · $100/month · 5 小时窗口: 5× Pro (relative-to-pro) · 来源: https://support.claude.com/en/articles/11049741-what-is-the-max-plan · 核实: 2026-09-17 - Max 5x · $100/month · 周限额: 未公布 (fixed-weekly) · 来源: https://support.claude.com/en/articles/11049741-what-is-the-max-plan · 核实: 2026-09-17 - Max 20x · $200/month · 5 小时窗口: 20× Pro (relative-to-pro) · 来源: https://support.claude.com/en/articles/11049741-what-is-the-max-plan · 核实: 2026-09-17 - Max 20x · $200/month · 周限额: 未公布 (discretion) · 来源: https://support.claude.com/en/articles/11049741-what-is-the-max-plan · 核实: 2026-09-17 - Team Standard seat · $25/seat/month ($20 billed annually) · 5 小时窗口: 未公布 (per-seat) · 来源: https://claude.com/pricing · 核实: 2026-09-17 - Team Standard seat · $25/seat/month ($20 billed annually) · 周限额: 未公布 (per-seat-weekly) · 来源: https://code.claude.com/docs/en/costs · 核实: 2026-09-17 - Team Premium seat · $125/seat/month ($100 billed annually) · 5 小时窗口: 5× Standard seat (relative-to-standard) · 来源: https://claude.com/pricing · 核实: 2026-09-17 - Team Premium seat · $125/seat/month ($100 billed annually) · 周限额: 未公布 (per-seat-weekly) · 来源: https://code.claude.com/docs/en/costs · 核实: 2026-09-17 厂商未公布的单元格写「未公布」，本站不估算。 改过什么？ - 2025-07-28: Anthropic：「我们将在 8 月底为 Claude Pro 与 Max 推出新的周限额」，叠加在原有的 5 小时会话限额之上；估计影响「不到 5% 的订阅用户」。 — https://x.com/AnthropicAI/status/1949898502688903593 窗口怎么算 订阅计划下的 Claude Code 与 claude.ai 共用一个额度池：「两个工具里的所有活动都计入同一份用量限额」。会话限额「每五小时重置」。claude.ai 的 Settings > Usage 显示 5 小时会话已用多少，「以及会话剩余的时间」；Claude Code 里的 /usage 显示同样的计划用量条。 周限额于 2025-07-28 宣布、2025 年 8 月底上线，估计影响「不到 5% 的订阅用户」。它「跨全部模型生效」，并「在分配给你账号的每周固定时间重置」；用量页还单独列出「仅 Opus」的周重置时间。启用用量积分（usage credits）时，5 小时限额的重置时间「不受影响」。 Team 席位按席位同样计算：每位成员的用量「从一份按 5 小时滚动窗口和周窗口重置的席位额度扣除」，并与 Claude 对话及 Cowork 共用；席位档位（Standard 或 Premium）决定额度大小。Anthropic 没有为任何档位公布消息条数或 token 数字，并保留酌情施加「周上限与月上限」的权利。 FAQ: Q: Claude Pro 或 Max 每 5 小时能发多少条消息，有公布的数字吗？ A: 没有。Anthropic 的 Pro 文章写道，消息条数「会随消息长度」以及所用模型或功能而变。Max 5x 与 Max 20x 只以 Pro 为基准定义：「Pro 计划每会话额度的五倍」和「20 倍」。claude.ai 的 Settings > Usage 与 Claude Code 的 /usage 显示你自己剩余的份额和会话剩余时间。 Q: Claude 的周限额什么时候重置？ A: 在分配给你账号的每周固定时间；Anthropic 不公布具体是哪天。用量页显示你的周限额「仅 Opus 与其他全部模型」各自的重置时间。触到 5 小时会话限额不会挪动周时钟，用量积分也不会改变 5 小时重置的时间。 Q: Claude Code 与 claude.ai、IDE 扩展共用额度吗？ A: 共用。「Pro 与 Max 计划的用量限额在 Claude 与 Claude Code 之间共享」，你在所有产品界面的用量都「计入同一份用量限额」。Team 计划的席位额度同样与 Claude 对话和 Cowork 共用。 Q: Anthropic 有没有为所有人重置过周限额？ A: 上方板面记录了 Anthropic 账号公告的每一次全员重置，每条都链接原帖，下方的规律条统计了次数。Anthropic 不会提前宣布重置，本站也不承诺下一次何时发生；板上的预计时间是上次重置加上中位间隔。 Sources: 1. What is the Pro plan? | Claude Help Center — https://support.claude.com/en/articles/8325606-what-is-the-pro-plan (verified 2026-09-17) 2. What is the Max plan? | Claude Help Center — https://support.claude.com/en/articles/11049741-what-is-the-max-plan (verified 2026-09-17) 3. Pricing | Claude (claude.com/pricing) — https://claude.com/pricing (verified 2026-09-17) 4. Manage costs effectively – Claude Code Docs — https://code.claude.com/docs/en/costs (verified 2026-09-17) 5. Usage limit best practices | Claude Help Center — https://support.claude.com/en/articles/9797557-usage-limit-best-practices (verified 2026-09-17) 6. Use Claude Code with your Pro or Max plan | Claude Help Center — https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan (verified 2026-09-17) 7. Anthropic on X, 2025-07-28: weekly rate limits for Pro and Max announced — https://x.com/AnthropicAI/status/1949898502688903593 (verified 2026-09-17) 8. Manage usage credits for paid Claude plans | Claude Help Center — https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans (verified 2026-09-17) 9. How do usage and length limits work? | Claude Help Center — https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work (verified 2026-09-17) ### Grok Bot usage limits URL: https://didreset.com/limits/grok Language: en · Published: 2026-09-18 · Updated: 2026-09-18 xAI publishes no numbers for Grok usage limits on any plan. What is documented is the shape: paid Grok plans share one weekly pool across Chat, Imagine, Voice and Build that "resets every week on a schedule shown in the Usage tab in Settings", and Grok Bot has its own pool where "Usage resets weekly". The docs describe no five-hour window. The last account-wide Grok reset on this board was announced {lastReset}. What are the limits on each plan? - Free · $0/month · 5-hour window: Not published (no-five-hour) · Source: https://docs.x.ai/grok/faq · Verified: 2026-09-17 - Free · $0/month · Weekly limit: Not published (free-tier) · Source: https://docs.x.ai/grok/faq · Verified: 2026-09-17 - SuperGrok · $30/month · 5-hour window: Not published (no-five-hour) · Source: https://docs.x.ai/grok/faq · Verified: 2026-09-17 - SuperGrok · $30/month · Weekly limit: Not published (weekly-pool) · Source: https://docs.x.ai/grok/faq · Verified: 2026-09-17 - SuperGrok Plus · $100/month · 5-hour window: Not published (no-five-hour) · Source: https://docs.x.ai/grok/faq · Verified: 2026-09-17 - SuperGrok Plus · $100/month · Weekly limit: Not published (plus-higher) · Source: https://x.ai/pricing · Verified: 2026-09-17 - SuperGrok Heavy · 5-hour window: Not published (no-five-hour) · Source: https://docs.x.ai/grok/faq · Verified: 2026-09-17 - SuperGrok Heavy · Weekly limit: Not published (price-not-captured) · Source: https://x.ai/pricing · Verified: 2026-09-17 - Grok Bot · included with SuperGrok, SuperGrok Plus, SuperGrok Heavy and paid Cursor plans · 5-hour window: Not published (no-five-hour) · Source: https://docs.x.ai/grok-bot/overview · Verified: 2026-09-17 - Grok Bot · included with SuperGrok, SuperGrok Plus, SuperGrok Heavy and paid Cursor plans · Weekly limit: Not published (bot-weekly) · Source: https://docs.x.ai/grok-bot/overview · Verified: 2026-09-17 Cells the vendor has not published are marked "Not published". This site does not estimate them. What has changed? - 2026-08-26: Grok Bot included with all SuperGrok, SuperGrok Plus, SuperGrok Heavy and paid Cursor plans, with "its own usage, separate from your Grok and Cursor plans"; @bot: "We're also resetting weekly usage limits for all users." — https://x.ai/news/grok-bot-more-plans - 2026-08-13: @grok resets limits for the Grok 4.6 launch and tells users to "Use a reset token from settings in Grok on desktop or mobile", the first official mention of a user-applied reset token for Grok. — https://x.com/grok/status/2087959772909949375 How the windows work For the Grok app, xAI’s FAQ describes a single pool: "Different products cost different amounts depending on how much compute that product requires", and "The usage pool limit resets every week on a schedule shown in the Usage tab in Settings". When it is met, "paid features will pause until your weekly limit resets", while the free-tier Chat and Voice limits stay available on their own schedule. Grok Bot is metered separately. Since 2026-08-26 it is included with SuperGrok, SuperGrok Plus, SuperGrok Heavy and paid Cursor plans and "comes with its own usage, separate from your Grok and Cursor plans". Its docs say "Usage resets weekly" and that subscriptions "include weekly usage; eligible accounts can add on-demand usage billed from model and token cost". If both a Cursor and a SuperGrok plan are linked, it "uses whichever has more usage". The docs point to two places to check: the Grok app shows the pool and its reset time under Settings → Usage, and Grok Bot’s troubleshooting page says to "review usage and billing from the account access page or Usage & Billing". Account-wide resets are announced on @grok and @bot; on 2026-08-13 @grok also told users to "Use a reset token from settings in Grok on desktop or mobile". FAQ: Q: Does Grok have a 5-hour window like Codex or Claude Code? A: xAI’s documentation describes one weekly pool for paid Grok plans and a separate weekly pool for Grok Bot; it does not describe a five-hour window. The table above therefore shows "Not published" in the five-hour column for every plan rather than a number this site made up. Your own reset time is in Settings → Usage in the Grok app. Q: How much usage does SuperGrok, SuperGrok Plus or SuperGrok Heavy include per week? A: xAI has not published a number. The pricing page describes SuperGrok as "Higher rate limits across all features" and SuperGrok Plus as "Significantly higher usage across Chat, Imagine, Voice & Build"; the Heavy tier’s price text was not captured in the snapshot read when this page was verified. The pool resets weekly on the schedule shown in Settings → Usage. Q: Does Grok Bot usage count against my SuperGrok or Cursor plan? A: No. xAI’s announcement says Grok Bot "comes with its own usage, separate from your Grok and Cursor plans, so anything you hand off to a Bot won't count against your existing usage". Its usage resets weekly, and if both plans are linked Grok Bot "uses whichever has more usage". Q: What does an official Grok reset look like? A: It is announced on @grok or @bot; for example on 2026-08-26, when Grok Bot was extended to all SuperGrok and Cursor Pro subscribers, @bot added "We're also resetting weekly usage limits for all users." Each reset on the board above links to its post. xAI does not announce resets in advance, and this site does not promise a time. Sources: 1. Pricing | xAI (x.ai/pricing, read via Wayback snapshot dated 2026-09-17) — https://x.ai/pricing (verified 2026-09-17) 2. FAQ – Grok Website / Apps | xAI Docs — https://docs.x.ai/grok/faq (verified 2026-09-17) 3. Grok Bot | xAI Docs (overview) — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 4. Frequently asked questions – Grok Bot | xAI Docs — https://docs.x.ai/grok-bot/faq (verified 2026-09-17) 5. Grok Bot is now included with more plans | xAI news, 2026-08-26 — https://x.ai/news/grok-bot-more-plans (verified 2026-09-17) 6. @grok on X, 2026-08-13: limits reset for the Grok 4.6 launch; "Use a reset token from settings" — https://x.com/grok/status/2087959772909949375 (verified 2026-09-17) 7. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 8. Troubleshooting – Grok Bot | xAI Docs — https://docs.x.ai/grok-bot/troubleshooting (verified 2026-09-17) ### Grok Bot 额度上限 URL: https://didreset.com/zh/limits/grok Language: zh · Published: 2026-09-18 · Updated: 2026-09-18 xAI 未公布 Grok 各计划的额度数字，只公布结构：付费计划共用跨 Chat、Imagine、Voice、Build 的周额度池，「每周按 Settings 里 Usage 标签的时间表重置」；Grok Bot 独立池「用量每周重置」；无 5 小时窗口。板上 Grok 上次全员重置公告于 {lastReset}。 各计划的额度是多少？ - Free · $0/month · 5 小时窗口: 未公布 (no-five-hour) · 来源: https://docs.x.ai/grok/faq · 核实: 2026-09-17 - Free · $0/month · 周限额: 未公布 (free-tier) · 来源: https://docs.x.ai/grok/faq · 核实: 2026-09-17 - SuperGrok · $30/month · 5 小时窗口: 未公布 (no-five-hour) · 来源: https://docs.x.ai/grok/faq · 核实: 2026-09-17 - SuperGrok · $30/month · 周限额: 未公布 (weekly-pool) · 来源: https://docs.x.ai/grok/faq · 核实: 2026-09-17 - SuperGrok Plus · $100/month · 5 小时窗口: 未公布 (no-five-hour) · 来源: https://docs.x.ai/grok/faq · 核实: 2026-09-17 - SuperGrok Plus · $100/month · 周限额: 未公布 (plus-higher) · 来源: https://x.ai/pricing · 核实: 2026-09-17 - SuperGrok Heavy · 5 小时窗口: 未公布 (no-five-hour) · 来源: https://docs.x.ai/grok/faq · 核实: 2026-09-17 - SuperGrok Heavy · 周限额: 未公布 (price-not-captured) · 来源: https://x.ai/pricing · 核实: 2026-09-17 - Grok Bot · included with SuperGrok, SuperGrok Plus, SuperGrok Heavy and paid Cursor plans · 5 小时窗口: 未公布 (no-five-hour) · 来源: https://docs.x.ai/grok-bot/overview · 核实: 2026-09-17 - Grok Bot · included with SuperGrok, SuperGrok Plus, SuperGrok Heavy and paid Cursor plans · 周限额: 未公布 (bot-weekly) · 来源: https://docs.x.ai/grok-bot/overview · 核实: 2026-09-17 厂商未公布的单元格写「未公布」，本站不估算。 改过什么？ - 2026-08-26: Grok Bot 随所有 SuperGrok、SuperGrok Plus、SuperGrok Heavy 与付费 Cursor 计划附带，「有自己的用量，与 Grok 和 Cursor 计划分开」；@bot：「我们同时为所有用户重置周用量限额。」 — https://x.ai/news/grok-bot-more-plans - 2026-08-13: @grok 为 Grok 4.6 发布重置限额，并告诉用户「在桌面或移动端 Grok 的设置里使用重置令牌」，这是官方首次提到 Grok 用户可自行使用的重置令牌。 — https://x.com/grok/status/2087959772909949375 窗口怎么算 对 Grok 应用，xAI 的 FAQ 描述的是一个池子：「不同产品消耗的额度不同，取决于该产品需要多少算力」，「用量池限额每周按 Settings 里 Usage 标签显示的时间表重置」。用满之后，「付费功能会暂停，直到你的周限额重置」，免费层的 Chat 与 Voice 限额仍按自己的时间表可用。 Grok Bot 单独计量。自 2026-08-26 起它随 SuperGrok、SuperGrok Plus、SuperGrok Heavy 与付费 Cursor 计划附带，「有自己的用量，与 Grok 和 Cursor 计划分开」。其文档写「用量每周重置」，订阅「包含每周用量；符合条件的账号可追加按模型与 token 成本计费的按需用量」。若同时绑定 Cursor 与 SuperGrok，它「使用用量更多的那一个」。 文档给了两个查看位置：Grok 应用在 Settings → Usage 显示额度池及其重置时间；Grok Bot 的故障排除页写「在账号访问页或 Usage & Billing 查看用量与账单」。全员重置由 @grok 与 @bot 公告；2026-08-13 @grok 还让用户「在桌面或移动端 Grok 的设置里使用重置令牌」。 FAQ: Q: Grok 有像 Codex 或 Claude Code 那样的 5 小时窗口吗？ A: xAI 的文档描述的是付费 Grok 计划的一个周额度池，以及 Grok Bot 单独的周额度池；没有描述 5 小时窗口。所以上表每个计划的 5 小时一列都写「未公布」，而不是本站编一个数字。你自己的重置时间在 Grok 应用的 Settings → Usage 里。 Q: SuperGrok、SuperGrok Plus、SuperGrok Heavy 每周包含多少用量？ A: xAI 没有公布数字。定价页把 SuperGrok 描述为「所有功能更高的速率限制」，把 SuperGrok Plus 描述为「Chat、Imagine、Voice 与 Build 的用量显著更高」；Heavy 档的价格文字在本页核实时读取的快照里没有渲染出来。额度池每周按 Settings → Usage 显示的时间表重置。 Q: Grok Bot 的用量会计入我的 SuperGrok 或 Cursor 计划吗？ A: 不会。xAI 的公告写道，Grok Bot「有自己的用量，与 Grok 和 Cursor 计划分开，所以交给 Bot 的任务不会计入你现有的用量」。它的用量每周重置；若两个计划都绑定，Grok Bot「使用用量更多的那一个」。 Q: Grok 的官方重置长什么样？ A: 由 @grok 或 @bot 公告；例如 2026-08-26 Grok Bot 扩展到所有 SuperGrok 与 Cursor Pro 订阅者时，@bot 补了一句「我们同时为所有用户重置周用量限额」。上方板面每一次重置都链接原帖。xAI 不会提前宣布重置，本站也不承诺时间。 Sources: 1. Pricing | xAI (x.ai/pricing, read via Wayback snapshot dated 2026-09-17) — https://x.ai/pricing (verified 2026-09-17) 2. FAQ – Grok Website / Apps | xAI Docs — https://docs.x.ai/grok/faq (verified 2026-09-17) 3. Grok Bot | xAI Docs (overview) — https://docs.x.ai/grok-bot/overview (verified 2026-09-17) 4. Frequently asked questions – Grok Bot | xAI Docs — https://docs.x.ai/grok-bot/faq (verified 2026-09-17) 5. Grok Bot is now included with more plans | xAI news, 2026-08-26 — https://x.ai/news/grok-bot-more-plans (verified 2026-09-17) 6. @grok on X, 2026-08-13: limits reset for the Grok 4.6 launch; "Use a reset token from settings" — https://x.com/grok/status/2087959772909949375 (verified 2026-09-17) 7. Grok Bot (@bot) on X, 2026-08-26: weekly usage limits reset for all users — https://x.com/bot/status/2092664183334846840 (verified 2026-09-17) 8. Troubleshooting – Grok Bot | xAI Docs — https://docs.x.ai/grok-bot/troubleshooting (verified 2026-09-17) ## Reset times by time zone (live data; every page converts the same record) - When does Codex reset?: https://didreset.com/when/codex - Codex resets in Beijing time (China Standard Time, UTC+8): https://didreset.com/when/codex/beijing - Codex resets in US Eastern time (New York, ET): https://didreset.com/when/codex/new-york - Codex resets in US Pacific time (Los Angeles, PT): https://didreset.com/when/codex/los-angeles - Codex resets in UK time (London, GMT/BST): https://didreset.com/when/codex/london - Codex resets in Central European time (Berlin, CET/CEST): https://didreset.com/when/codex/berlin - Codex resets in Japan time (Tokyo, JST, UTC+9): https://didreset.com/when/codex/tokyo - Codex resets in Singapore time (SGT, UTC+8): https://didreset.com/when/codex/singapore - Codex resets in India time (IST, UTC+5:30): https://didreset.com/when/codex/india - Codex resets in Australian Eastern time (Sydney, AEST/AEDT): https://didreset.com/when/codex/sydney - Codex resets in UTC (Coordinated Universal Time): https://didreset.com/when/codex/utc - Codex 什么时候重置？: https://didreset.com/zh/when/codex - Codex 重置时间 · 北京时间（中国标准时间，UTC+8）: https://didreset.com/zh/when/codex/beijing - Codex 重置时间 · 美东时间（纽约，ET）: https://didreset.com/zh/when/codex/new-york - Codex 重置时间 · 美西时间（洛杉矶，PT）: https://didreset.com/zh/when/codex/los-angeles - Codex 重置时间 · 英国时间（伦敦，GMT/BST）: https://didreset.com/zh/when/codex/london - Codex 重置时间 · 欧洲中部时间（柏林，CET/CEST）: https://didreset.com/zh/when/codex/berlin - Codex 重置时间 · 日本时间（东京，JST，UTC+9）: https://didreset.com/zh/when/codex/tokyo - Codex 重置时间 · 新加坡时间（SGT，UTC+8）: https://didreset.com/zh/when/codex/singapore - Codex 重置时间 · 印度时间（IST，UTC+5:30）: https://didreset.com/zh/when/codex/india - Codex 重置时间 · 澳洲东部时间（悉尼，AEST/AEDT）: https://didreset.com/zh/when/codex/sydney - Codex 重置时间 · UTC（协调世界时）: https://didreset.com/zh/when/codex/utc - When does Claude Code reset?: https://didreset.com/when/claude - Claude Code resets in Beijing time (China Standard Time, UTC+8): https://didreset.com/when/claude/beijing - Claude Code resets in US Eastern time (New York, ET): https://didreset.com/when/claude/new-york - Claude Code resets in US Pacific time (Los Angeles, PT): https://didreset.com/when/claude/los-angeles - Claude Code resets in UK time (London, GMT/BST): https://didreset.com/when/claude/london - Claude Code resets in Central European time (Berlin, CET/CEST): https://didreset.com/when/claude/berlin - Claude Code resets in Japan time (Tokyo, JST, UTC+9): https://didreset.com/when/claude/tokyo - Claude Code resets in Singapore time (SGT, UTC+8): https://didreset.com/when/claude/singapore - Claude Code resets in India time (IST, UTC+5:30): https://didreset.com/when/claude/india - Claude Code resets in Australian Eastern time (Sydney, AEST/AEDT): https://didreset.com/when/claude/sydney - Claude Code resets in UTC (Coordinated Universal Time): https://didreset.com/when/claude/utc - Claude Code 什么时候重置？: https://didreset.com/zh/when/claude - Claude Code 重置时间 · 北京时间（中国标准时间，UTC+8）: https://didreset.com/zh/when/claude/beijing - Claude Code 重置时间 · 美东时间（纽约，ET）: https://didreset.com/zh/when/claude/new-york - Claude Code 重置时间 · 美西时间（洛杉矶，PT）: https://didreset.com/zh/when/claude/los-angeles - Claude Code 重置时间 · 英国时间（伦敦，GMT/BST）: https://didreset.com/zh/when/claude/london - Claude Code 重置时间 · 欧洲中部时间（柏林，CET/CEST）: https://didreset.com/zh/when/claude/berlin - Claude Code 重置时间 · 日本时间（东京，JST，UTC+9）: https://didreset.com/zh/when/claude/tokyo - Claude Code 重置时间 · 新加坡时间（SGT，UTC+8）: https://didreset.com/zh/when/claude/singapore - Claude Code 重置时间 · 印度时间（IST，UTC+5:30）: https://didreset.com/zh/when/claude/india - Claude Code 重置时间 · 澳洲东部时间（悉尼，AEST/AEDT）: https://didreset.com/zh/when/claude/sydney - Claude Code 重置时间 · UTC（协调世界时）: https://didreset.com/zh/when/claude/utc - When does Grok Bot reset?: https://didreset.com/when/grok - Grok Bot resets in Beijing time (China Standard Time, UTC+8): https://didreset.com/when/grok/beijing - Grok Bot resets in US Eastern time (New York, ET): https://didreset.com/when/grok/new-york - Grok Bot resets in US Pacific time (Los Angeles, PT): https://didreset.com/when/grok/los-angeles - Grok Bot resets in UK time (London, GMT/BST): https://didreset.com/when/grok/london - Grok Bot resets in Central European time (Berlin, CET/CEST): https://didreset.com/when/grok/berlin - Grok Bot resets in Japan time (Tokyo, JST, UTC+9): https://didreset.com/when/grok/tokyo - Grok Bot resets in Singapore time (SGT, UTC+8): https://didreset.com/when/grok/singapore - Grok Bot resets in India time (IST, UTC+5:30): https://didreset.com/when/grok/india - Grok Bot resets in Australian Eastern time (Sydney, AEST/AEDT): https://didreset.com/when/grok/sydney - Grok Bot resets in UTC (Coordinated Universal Time): https://didreset.com/when/grok/utc - Grok Bot 什么时候重置？: https://didreset.com/zh/when/grok - Grok Bot 重置时间 · 北京时间（中国标准时间，UTC+8）: https://didreset.com/zh/when/grok/beijing - Grok Bot 重置时间 · 美东时间（纽约，ET）: https://didreset.com/zh/when/grok/new-york - Grok Bot 重置时间 · 美西时间（洛杉矶，PT）: https://didreset.com/zh/when/grok/los-angeles - Grok Bot 重置时间 · 英国时间（伦敦，GMT/BST）: https://didreset.com/zh/when/grok/london - Grok Bot 重置时间 · 欧洲中部时间（柏林，CET/CEST）: https://didreset.com/zh/when/grok/berlin - Grok Bot 重置时间 · 日本时间（东京，JST，UTC+9）: https://didreset.com/zh/when/grok/tokyo - Grok Bot 重置时间 · 新加坡时间（SGT，UTC+8）: https://didreset.com/zh/when/grok/singapore - Grok Bot 重置时间 · 印度时间（IST，UTC+5:30）: https://didreset.com/zh/when/grok/india - Grok Bot 重置时间 · 澳洲东部时间（悉尼，AEST/AEDT）: https://didreset.com/zh/when/grok/sydney - Grok Bot 重置时间 · UTC（协调世界时）: https://didreset.com/zh/when/grok/utc