How do I check my Codex usage and remaining limit?

Last updated: 26 September 2026 · Deutsch

Where you see your usage

• Codex CLI. Type /status in a running session. It shows the active model and settings, and your remaining limits. /usage shows your token activity by day or week and lets you use a saved reset. With /statusline you can put the limits into the footer, so they stay in view.
• IDE extension. /status in the Codex panel shows the chat, its context use and your rate limits.
• Web and desktop app. The usage page at chatgpt.com/codex/settings/usage, or Settings → Usage in the desktop app. OpenAI points here for current limits and reset times.

One trap: /status also shows how much context is left. That is the chat’s working memory, not your plan’s limit. A full context means “start a new chat”, not “wait for a reset”.

Two windows: five hours and one week

Depending on your plan, Codex counts your use against a window of about five hours and a weekly window. Where both apply, you need room in both. You can hit the five-hour limit with weekly room to spare, and the weekly limit stops you even when the five-hour window is fresh.

Local tasks in the CLI or IDE, cloud tasks on the web, and ChatGPT Work inside ChatGPT all draw on the same allowance. There is no separate pot per surface.

Not every account shows both windows at all times. Some accounts see only the weekly window for a while, even in /status. If you see just one window, that is what OpenAI sends for your account right now.

How much fits in a window

OpenAI names no fixed number. Its pricing page lists wide ranges per model and plan, and says plainly that these are estimates, not fixed message limits.

What a message costs depends on the model, how much context it carries, reasoning effort, tool calls and caching. Fast mode and image generation use the allowance faster. Pro offers five or twenty times the Plus usage. Enterprise and Edu plans with flexible pricing have no fixed rate limits; use there is paid in credits.

When the windows reset

Each window has its own reset time. The usage page and /status show it. There is no weekday that applies to everyone: the date depends on your own window, and it moves when you use a reset.

• Saved (“banked”) resets sit in your account until you apply them or they expire. A full one refreshes both windows and moves your weekly date. OpenAI’s example: your week was due to reset on Friday, you apply a reset and carry on on Tuesday — the next weekly reset comes around the following Tuesday, and Friday’s is gone.
• Bought resets are offered to some Plus and Pro personal accounts, on the web and in the desktop app. They take effect at once, even if you still have room, cannot be saved for later, and also move your weekly date. OpenAI shows the price at checkout and names none in its help pages.

What happens at the limit

A task that is already running may finish its current turn, within fair use. After that, you can add credits, apply a saved reset, buy a reset, upgrade, or wait for the window to reset.

Which of these you get depends on the plan; the limit banner and the usage page show your options. Business, Edu and Enterprise workspaces with flexible pricing buy workspace credits instead. And any user can run extra local tasks with an API key, at API prices.

What credits are

Credits pay for use beyond what your plan includes. Your plan’s allowance is used first; only then does Codex draw on credits. OpenAI lists how many credits each model uses per million tokens; the price of a credit depends on your plan or contract.

Credits are not cash. They sit apart from your ChatGPT wallet, and OpenAI does not refund them unless the law requires it. Don’t confuse them with reset vouchers either — how AI Cockpit keeps the two apart is explained in Codex quotas and reset times.

Plan allowance or API costs

How you sign in to Codex decides how you pay:

• Signed in with ChatGPT — your plan’s allowance, the two windows, and credits once they run out. A flat monthly fee, with limits.
• Signed in with an API key — pay per use, with no cloud features such as GitHub code review or Slack.

Codex with an API key does not count against your ChatGPT plan. You pay per token at API prices, billed to your OpenAI API account. The plan’s windows and credits do not apply.

So check API spend in your API account, not on the Codex usage page. How AI Cockpit reads API costs is covered in AI API costs at a glance.

In AI Cockpit

The ChatGPT/Codex card sits in the Mac menu bar, in the app on iPhone and iPad, and on Apple Watch (via the iPhone app) with one window per page. You sign in once inside the app; there is no API key for this card.

• The five-hour and the seven-day window, each with the share used and its reset time. If OpenAI sends only the weekly window, the card shows only that one.
• A warning colour by default at 75 and 90 per cent on the Mac, at 80 and 95 on iPhone and Apple Watch. You can move the thresholds in Settings.
• With history turned on, a forecast of when a window will be full at your current pace. If you allow notifications, it warns you before the five-hour window fills.
• Your credit balance, the credits spent since the start of the month, and whether a free window reset is waiting, with its expiry date.

What it does not do: buy credits or resets, split the limit by model, or show tokens.

What the card reads and where the figures come from is in Codex quotas and reset times. Short answers to other questions are on the FAQ page.