The best Codex usage tracker for most developers is UsageAtlas, a free and open-source desktop app that shows your Codex session and weekly windows, remaining capacity, reset times, tokens and estimated cost on your own computer. Codex's built-in /status command is still the quickest check inside a running session, and the two work well together.
What a Codex usage tracker should show #
Codex limits come in a five-hour window, and weekly limits may also apply, according to OpenAI's pricing page. That makes a single number less useful than a picture over time. A good tracker answers four questions:
- How much room is left right now , in each window, before a long refactor or test run.
- When each window resets , so you know whether to wait or switch tools.
- What the work used : tokens, requests and models from your own sessions.
- How this week compares to last week, which a one-off check cannot tell you.
How UsageAtlas tracks Codex #
UsageAtlas connects to the Codex CLI you are already signed in to on this machine. From there it shows the supported session and weekly windows, remaining capacity and reset times, plus plan and credit details when Codex reports them. Next to that it reads local session activity and turns it into tokens, requests, models and an estimated cost.
The app keeps a daily view and a history view, a model mix, and time-of-day and weekday patterns. You can set a threshold for a usage window and get one desktop alert per reset cycle when you cross it, which is the part a manual check cannot do for you.
It is not only for Codex. Claude Code, Cursor, OpenCode, GitHub Copilot and other coding tools can sit in the same view, and a per-tool filter narrows it to Codex when that is all you want to see.
Tracker vs built-in checks #
| Need | /status in Codex |
OpenAI usage page | UsageAtlas |
|---|---|---|---|
| Remaining limits now | Yes | Yes | Yes |
| Reset times | In the session | Yes | Yes |
| Local token history by day | No | No | Yes |
| Estimated cost | No | No | Yes |
| Threshold alerts | No | No | Yes |
| Other coding tools | No | No | Yes |
OpenAI's docs describe /status as showing the chat ID, context usage and rate limits, and /usage as opening usage and billing settings when you sign in with ChatGPT. The web dashboard lives at chatgpt.com/codex/settings/usage.
Set it up #
- Download UsageAtlas for Windows, macOS or Linux from the UsageAtlas download page .
- Open it while the Codex CLI is signed in on the same computer.
- Check the Today view for current windows, then History once a few days have collected.
- Set an alert threshold on the weekly window if you tend to run into it late in the week.
No account is needed for local tracking. A free Google or GitHub sign-in is optional and only saves usage summaries so you can restore them on another machine. Prompts and responses are not read.
Limits worth knowing #
Cost figures are estimates for planning, not an invoice. Credits only appear when Codex reports them, and provider data can arrive late or change. For a purchase or billing question, confirm in your OpenAI account.
When /status is enough #
If you only use Codex, rarely get near a limit and do not care about history, /status inside the CLI covers you. A tracker earns its place once you start planning work around the weekly window, or once Codex is one of several tools you pay for.
Who it suits #
UsageAtlas fits developers who keep Codex running for long sessions, switch between Codex and Claude Code during the week, or want a record of which models took most of their tokens. Because it runs locally, it also suits people who would rather not send session data to another hosted service.