Does reasoning effort count toward the Codex usage limit?
Updated
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.
What the official docs say
Usage depends on the model, where the task runs, task complexity, context, reasoning, speed, and tools.
Quoted from OpenAI Help Center: Using Codex with your ChatGPT plan Higher effort can use more of your allowance and does not always produce a better result.
Quoted from OpenAI Help Center: 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.
Quoted from OpenAI Help Center: 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.
Quoted from Codex docs: Pricing (learn.chatgpt.com) Fast mode applies a 2.5x multiplier to Astra's Standard rate.
Quoted from Codex docs: Pricing (learn.chatgpt.com) We made it too easy to use the highest-compute settings without making the impact on usage limits sufficiently clear.
Quoted from Tibo Sottiaux on X, 2026-07-10 (highest-compute settings and usage limits)
How it is counted
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.'
| 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
Does a lower reasoning level guarantee less usage?
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.
Does Fast mode use more of my Codex allowance?
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.'
Did OpenAI acknowledge that high-compute settings drained limits?
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.
How do I see how much a task used?
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
- OpenAI Help Center: Managing usage with GPT-6 Astra in Work and Codexverified 2026-09-17
- Codex docs: Pricing (learn.chatgpt.com)verified 2026-09-17
- OpenAI Help Center: Using Codex with your ChatGPT planverified 2026-09-17
- Tibo Sottiaux on X, 2026-07-10 (highest-compute settings and usage limits)verified 2026-09-17
Want the next Codex reset on your phone?