Skip to content
Burnstate

[guides] codex

Codex usage limits: the 5-hour and weekly windows

Codex comes with your ChatGPT plan, and like Claude Code it meters work on clocks instead of per token. What you get depends on the plan, the model and where the task runs. Here is how the limits fit together.

Updated 6 min read

The 5-hour and weekly windows

On ChatGPT Plus and standard Business seats, Codex usage counts against two windows at once: a rolling 5-hour window and a weekly one. When either runs out, Codex stops until that window resets, even if the other has room.

As of October 2026, Pro plans have no 5-hour limit. They keep a weekly allowance, so a Pro user can work as hard as they like in a single afternoon, as long as the week holds out.

What each plan gets

Plus
$20 a month
A 5-hour and a weekly window.
Pro
$100, $200 or $500 a month
A weekly allowance and no 5-hour window.
Business
Per seat
Plus-level limits per seat, or workspace credits with flexible pricing.
Enterprise and Edu
Contact sales
No fixed limits with flexible pricing; Plus-level limits per seat without it.

OpenAI gives the 5-hour allowance as a message range per model. On Plus, the lightest model fits hundreds to thousands of local messages in a window, the most capable one a few dozen at most. The ranges are wide because a message that reads a big repository costs far more than a one-line fix.

Local and cloud tasks

Codex runs in two places: locally, in the CLI or your editor, on your machine; and in the cloud, where a task gets its own environment and works on its own. Both draw from the same allowance.

Cloud tasks can use more of it, because they run longer and set up their own environment. Code review through GitHub counts separately.

How to check what’s left

  • In the Codex CLI: run /status for the current windows and when they reset.
  • In ChatGPT: Codex settings › Usage, at chatgpt.com/codex/settings/usage, shows the 5-hour and weekly windows.

Both show what you’ve used, not whether it lasts. To know that, compare usage with how much of the window has passed; see how to pace a limit, or enter both numbers in the pace calculator.

When you run out

  • Wait. The 5-hour window comes back first; the weekly one takes longer.
  • Buy credits. Plus and Pro can buy credits that keep Codex going past the limits. Business, Enterprise and Edu workspaces with flexible pricing buy them for the workspace.
  • Use an API key. Sign Codex in with an OpenAI API key and you pay API rates instead, with no plan windows.
Credits are spent per token, at a rate that depends on the model. If you buy credits most weeks, compare that spend with the next plan up.

Codex in Burnstate

Burnstate reads your Codex session and weekly limits with the Codex CLI’s login, or a ChatGPT sign-in in your browser. Each limit gets a bar with a white tick for the clock, so you see at a glance whether you’re ahead of it, and when you’d run out at your pace. See Connect providers and Limits and pace.

Sources

Plans and limits change often. These are the pages this guide was checked against.

  1. Codex pricing and limits
  2. Weekly work and Codex rate limit resets

[burnstate]

See every limit against the clock, in your menu bar.

Burnstate reads your Claude Code, Codex, Cursor and Vercel limits, marks how much of each window has passed, and tells you when you’d run out at your current pace.