The quick answer
For Codex with ChatGPT sign-in, open the official usage dashboard using the same account and workspace as your task. In the CLI, use /status to inspect the session and any displayed limits. Read each limit's label, percentage and reset time together; token activity and context are different measures.
Where can I see my remaining Codex allowance?
Start with the dashboard, then compare the account and workspace with your client. The authentication guide distinguishes ChatGPT sign-in from API-key access. In the CLI, codex login status helps identify the authentication method; it is not a remaining-quota report.
If you use an API key, inspect the relevant API project and organization instead. A ChatGPT subscription dashboard does not establish that project's balance or rate limits. A custom provider can have different metering again.
/status and /usage do different things
The developer command reference documents CLI /status for session information. Check it alongside any available limit display for the session you are actually running.
The developer command guide describes CLI /usage as a menu for token activity and available earned resets; /usage daily, /usage weekly and /usage cumulative show token activity. These totals do not establish your remaining subscription percentage. In the desktop app, /usage opens usage and billing settings. Check the interface before transferring a command's meaning between clients.
Does the percentage mean used or remaining?
Read the label before doing arithmetic. In a hypothetical display, 20% used leaves 80% of that window; 20% remaining leaves 20%. Neither tells you exactly how many future tasks will fit, because task consumption varies.
Record each displayed window separately. The current plan documentation says Pro plans have no five-hour limit; do not expect identical windows across plans. An exhausted weekly allowance can still matter when another displayed window has recovered. Use the reset-time guide for interpreting the applicable recovery time.
Context, tokens, credits and allowance
- Context: information available within a conversation, not the account's remaining allowance.
- Token activity: recorded model processing, not a formula for included-plan percentages.
- Credits: a balance for eligible paid usage, distinct from included allowance.
- Subscription limits: the restrictions and windows shown for the signed-in account. ChatGPT Work and Codex share usage.
For consumption factors such as model choice and tools, see how Codex usage works.
Missing or inconsistent usage information
Confirm account, workspace, sign-in method, client version and the time each screen was checked. Other activity on the account can explain different readings. If a field is absent, keep it unknown rather than estimating it from token totals or another user's screenshot.
If usage is exhausted, follow limit recovery steps. If the error is 429, use the 429 diagnosis before assuming exhaustion. Save only an explicit reset time in the personal timer; this site does not read your allowance or verify recovery.