Last verified:
Codex usage limit: how much you get, how to check it, and what to do when you hit it
Codex (CLI, IDE extension, web and desktop) shares one usage allowance with your ChatGPT plan. On Plus and Standard Business it is measured over a five-hour window plus a weekly limit; Pro plans currently have no five-hour limit but still have a weekly allowance.
OpenAI does not publish a fixed message count; the numbers below are its own estimated ranges. Check Settings → Usage or run /status to see what you have left.
What are the Codex usage limits per plan?
Estimated local messages per five-hour window for Plus and Standard Business, from OpenAI's Codex pricing page as read on Oct 29, 2026:
| Model | Plus | Standard Business |
|---|---|---|
| GPT-6 Astra | 5–45 | 5–45 |
| GPT-6.1 Sol | 15–160 | 15–160 |
| GPT-6.1 Terra | 35–285 | 35–285 |
| GPT-6 Sol | 15–150 | 15–150 |
| GPT-6 Luna | 350–3,000 | 350–3,000 |
- These are estimates, not guaranteed counts. Cloud tasks can use more than local messages, and weekly limits may also apply.
- Pro 100 / Pro 200 / Pro 500: no five-hour limit right now, but an included weekly allowance (OpenAI Help).
- Free and Go: Codex is included with limits that vary by plan; Go uses a monthly allowance (resets article). Codex Cloud is not included with Free or Go (OpenAI Help).
- Enterprise/Edu with flexible pricing: no fixed rate limits; usage scales with credits.
Limits change often. If these numbers differ from what you see, trust the official pricing page and your dashboard.
Codex limits compared by plan
Only what OpenAI states officially (Oct 29, 2026):
| Plan | Price | Five-hour window | Weekly / period limit |
|---|---|---|---|
| Free, Go | Free / Go price | Varies by plan | Go: monthly allowance; no Codex Cloud |
| Plus | $20/month | Yes (ranges above) | Yes, weekly may apply |
| Pro 100 / 200 / 500 | $100 / $200 / $500 per month | None currently | Yes, included weekly allowance |
| Business Standard | Per seat | Yes (same ranges as Plus) | Yes, weekly may apply |
| Enterprise/Edu (flexible pricing) | Credits | No fixed limit | Scales with credits |
OpenAI does not publish exact weekly numbers. See the pricing page.
How do I check my remaining Codex usage?
In ChatGPT
Some users found that the old ring widget showing the five-hour and weekly percentages is gone. The official place to check now is Settings → Usage (the usage dashboard). It shows which allowance (five-hour or weekly) is used up, your credit balance, and the reset time.
In the Codex CLI
In an active session, enter:
/status
If the percentage jumps around (for example 93% → 33% → 0%), wait a minute and refresh. Recent usage can post late. If the drop is real and unexplained, record it (see below).
What should I do when I hit the Codex usage limit or rate limit?
If you hit the limit during an active turn, OpenAI says the agent can usually finish that turn. After that, your options are:
Wait for the reset
The five-hour window restarts with your first message after it ends. The weekly limit resets at the end of the weekly period. Both times are shown in Settings → Usage. Where both limits apply, you need allowance left in both.
Switch to a lighter model
Lighter models stretch the same allowance much further (compare the ranges above). Turn off Fast or Ultrafast mode: Fast uses included usage at about 2.5x and Ultrafast at about 8x the Standard rate.
Lower the reasoning level
Reasoning is one of the factors OpenAI lists as driving usage. High or extra-high reasoning effort on every request uses up the five-hour window quickly. Use a lower level for routine edits and save high effort for hard problems.
Buy credits
Plus and Pro users can buy additional credits to keep working without upgrading. Business/Edu/Enterprise with flexible pricing can buy workspace credits.
Use an API key
All users can run extra local chats with an OpenAI API key, billed at standard API rates. API rate limits then apply instead of plan limits.
Upgrade your plan
Pro plans remove the five-hour window. Compare plans on the pricing page.
Why does my Codex usage run out so fast?
A common complaint from Plus users: the five-hour window runs out after about 1.5 to 2 hours of heavy work, and the weekly quota is gone in two days. That is expected with large contexts, long agent runs and high reasoning effort, because cost scales with work done, not prompts sent.
Usage is not counted per prompt. Model, context size, reasoning, tool use, retrieval, caching and speed mode all change how much each message costs, and long-running or large-context tasks use far more. Codex usage is also shared with other agentic features on your plan (such as ChatGPT Work and ChatGPT for Excel).
Separately, some users report sudden, unexplained jumps. In openai/codex issue #54677 (opened Oct 16, 2026, still open), a Windows desktop user saw the remaining weekly allowance drop from 77% to 10% in 38 seconds while local logs showed only ~381k tokens, mostly cached input. If you see something similar, note timestamps and percentages and add them to that issue.
Sources
FAQ
What is the Codex usage limit on ChatGPT Plus?
There is no fixed message count. OpenAI estimates a range of local messages per five-hour window that depends on the model (for example 15–160 for GPT-6.1 Sol and 5–45 for GPT-6 Astra as of Oct 29, 2026), and a weekly limit may also apply. Check the official pricing page for current numbers.
Do Codex Pro plans have a usage limit?
Pro 100, Pro 200 and Pro 500 currently have no five-hour limit in Codex, but they still have an included weekly allowance. Check Settings → Usage for what remains.
How do I check my remaining Codex usage?
Open Settings → Usage (the usage dashboard) in ChatGPT, or type /status in an active Codex CLI session. Both show remaining allowance and reset times.
When does the Codex limit reset?
The five-hour window starts with your first message after the previous window ends; the weekly limit resets at the end of the weekly usage period. The exact reset time is shown in Settings → Usage.
Is 'codex rate limit' the same as the usage limit?
When signed in with ChatGPT, hitting the rate limit usually means your five-hour or weekly plan allowance is used up. With an API key, rate limits are the standard API rate limits for your organization and are billed separately.
Why did my weekly Codex quota drop so fast?
Usage depends on model, context size, reasoning, tools and speed mode; Fast mode uses included usage about 2.5x faster. Some users have also reported unexplained sudden drops, tracked in GitHub issue #54677.
Can I keep using Codex after hitting the limit?
Yes: wait for the reset, switch to a lighter model, buy additional credits (Plus and Pro), or run local chats with an API key at standard API rates.
My limit reset but a third-party client still says I'm limited. Why?
Some third-party coding clients cache the limit state or use their own quotas. Check Settings → Usage in ChatGPT first; if it shows allowance available, restart or re-authenticate the third-party client, and contact its support if it persists.
Does lowering reasoning effort save Codex usage?
Yes. OpenAI lists reasoning as one of the factors that determine usage, so a lower reasoning level for routine tasks makes the allowance last longer.
This is an independent, unofficial help site. It is not affiliated with or endorsed by OpenAI. Plans, limits and versions change; always check the official pages linked here.