Codex usage limits: how OpenAI meters your plan
A five-hour window, possible weekly limits and a shared allowance, and why OpenAI publishes ranges instead of fixed numbers.
By ClaudeGPT Si Newsroom, Editorial team··2 min read
Codex usage limits are measured over time, not per prompt: on Plus and Standard Business, OpenAI meters use in five-hour periods, and weekly limits may apply on top. The cleanest way to see where you stand is the official usage dashboard, plus a tracker such as UsageAtlas if you want the history and an early warning.
We do not quote allowances here. OpenAI revises them, and its own page calls them estimates.
The basic shape
Codex comes with every ChatGPT plan: Free, Go, Plus, Pro, Business, Edu and Enterprise. On a paid plan the allowance refills on a schedule rather than being a fixed monthly bucket.
- Five-hour periods. OpenAI's estimates for Plus and Standard Business are given per five-hour period.
- Pro. Pro plans currently have no five-hour limit.
- Weekly limits. OpenAI says weekly limits "may also apply," so a long week of heavy work can stop you even when the short window has room.
- One pool. Local messages and cloud chats share the same allowance, and ChatGPT Work draws on it too.
Why there is no fixed number
OpenAI is direct about this: the estimates "are not fixed message limits." What a message costs depends on several things at once.
| Factor | Effect on your allowance |
|---|---|
| Model choice | Each model draws at its own rate |
| Task size and complexity | Long, multi-step work uses more |
| Context held | Sessions holding more context use more |
| Local or cloud | Cloud tasks may use more |
| Reasoning, tools, retrieval | All change the cost |
That is why two prompts that look alike can draw very different amounts. Prompt length on its own is a poor guide.
What happens at the cap
If you hit the limit in the middle of a turn, Codex lets that turn finish, subject to fair use. After that you have three paths:
- Wait for the window to reset.
- On Plus or Pro, buy credits and carry on without changing plans. Business, Edu and Enterprise workspaces on flexible pricing buy workspace credits.
- Run extra local chats with an API key, billed at standard API rates.
API token prices are separate from subscription usage, so do not use them to work out your plan allowance.
Enterprise and Edu workspaces with flexible pricing have no fixed rate limits; usage scales with credits. Those without flexible pricing get the same per-seat limits as Plus for most features.
Where to read your limits
/statusin the Codex CLI shows your rate limits, along with the chat ID and context usage.- The usage dashboard at
chatgpt.com/codex/settings/usageshows current limits and reset times./usageopens it when you sign in with ChatGPT. - UsageAtlas keeps the picture on your desktop: session and weekly windows, remaining capacity, credits where reported, and a daily history.
The last one is our pick for anyone who uses Codex every day. It is free and open source, runs on Windows, macOS and Linux, needs no account for local tracking and does not read prompts or responses. It reads usage through the signed-in Codex CLI on your machine and can send one desktop alert per window at a threshold you choose. Its cost numbers are estimates, and provider data can lag, so for billing questions OpenAI's settings stay the source of truth.
If you only open Codex now and then, the dashboard alone is enough.