GPT-6 Astra 为什么把我的 Codex 5 小时额度用得这么快？

公布的估算把差距摆得很明白。帮助中心「estimated local messages per five-hour period」那张表：GPT-6 Astra 在 Plus 是 5–45、Pro 5x 是 25–225、Pro 20x 是 100–900；同样的计划上 GPT-5.6 Sol 是 10–100 / 50–500 / 200–2,000，GPT-5.6 Luna 是 250–2,000 / 1,250–10,000 / 5,000–40,000。「These are not fixed message limits. Actual usage varies by task, model and settings, and weekly limits may also apply.」token 费率卡上，Astra 每百万输入 token 250 credits、缓存输入 25、输出 1,250；Sol 是 100 / 10 / 500，Luna 是 5 / 0.5 / 30。

设置会把模型成本再乘上去。「Different models can use different amounts of your allowance for the same task. Larger inputs and outputs, higher reasoning settings, Fast mode and tasks with multiple steps can also increase usage.」「Higher effort can use more of your allowance and does not always produce a better result」，而且「Fast mode applies a 2.5x multiplier to Astra's Standard rate.」定价页补充：「larger projects, long-running tasks, or extended sessions that require the agent to hold more context will use significantly more per message.」

在 CLI 里能改什么：/model「Choose the active model (and reasoning effort, when available)」；/fast 关掉 Fast 档；/compact「Summarize the visible chat to free tokens」或 /new 开新对话；前后都用 /status 看「5h limit」进度条。OpenAI 对 effort 的建议：「Astra at Low effort can outperform Sol at High effort. If you've been getting good results with Sol at High, try Astra at Low or Medium as a starting point.」还有一句提醒，光切换不是补额度：「Switching models does not restore allowance in a shared usage pool.」issue #42987（2026-09-05）是一条用户报告：Plus 的 5 小时额度在两轮 Astra Medium 里从 100% 到 42% 再到 0%；把它当作一次观察，不是规则。