ChatGPT & Codex usage limits explained (and how to track them)
How ChatGPT Plus/Pro and OpenAI Codex usage limits work, what resets when, and how to monitor your remaining usage across accounts.
OpenAI's Codex (the coding agent behind the Codex CLI and IDE extensions) draws on your ChatGPT plan, and like Claude it enforces usage limits that aren't shown as a clear meter. Here's what they are and how to keep an eye on them.
Where Codex usage comes from
Codex uses your ChatGPT subscription (Plus, Pro, or Business). Usage is metered on a rolling weekly window, with separate buckets for standard models and the heavier GPT-5.x-Codex models. Run a lot of long agentic coding sessions and you'll feel the weekly cap first.
Why tracking it is hard
There's no live percentage in the Codex CLI, and if you use more than one ChatGPT account you can't see them together. The credentials live in `~/.codex/auth.json`, but reading raw tokens tells you nothing about how much you have left.
Track it automatically
How Much Codex reads the Codex login already on your machine (or a pasted `~/.codex/auth.json`) and shows your weekly usage per model, per account, with reset countdowns and warnings — so you stop getting blocked mid-review. Your first account is free.
Frequently asked questions
- Does Codex have a weekly limit?
- Yes — Codex usage is metered weekly against your ChatGPT plan, with separate buckets for standard and GPT-5.x-Codex models.
- Can I check my ChatGPT/Codex usage across multiple accounts?
- Not natively. How Much Codex shows every connected ChatGPT/Codex account's weekly usage on one dashboard.