Codex CLI
Codex CLI usage limit: windows, reset and how to see it
Signed in to the Codex CLI with ChatGPT, you have a usage allowance measured over time windows, each with its own reset time. Plans have had a rolling five-hour window and a weekly window; some have a single longer one, and OpenAI changes them, so check the Codex pricing page. The limits belong to your account, not a session: every Codex session on every machine draws on the same windows. Type /status after a reply to see yours.
Three ways to see it
/statusin a Codex session. Shows how much of each window you used and when it resets. Send one prompt first; Codex learns the numbers from its replies.- The usage page at chatgpt.com/codex/settings/usage. Works without a running session.
- The session logs. After each reply Codex writes a
rate_limitsobject under~/.codex/sessions. The full guide shows the format and a command to read it.
Keep it in the VS Code status bar
The free Claude Code & Codex Usage Meter reads those log entries and shows Codex's windows in the status bar, next to Claude Code's, and warns before you run out. Paneherd shows them in its sidebar. Neither sends anything anywhere.
Using an API key instead
With an API key you pay per token and the plan windows do not apply; the API's own rate limits do.
Frequently asked questions
Do more Codex sessions use the limit faster?
Yes. Three sessions use a five-hour window three times as fast.
Is Codex's limit the same as Claude Code's?
No. Each tool bills its own plan, which is why spreading work across agents can stretch your day. For Claude Code, see Claude Code usage limit.