Codex (OpenAI)
- Type /status in the Codex CLI during an active session.
- Read the row for your window: percentage left and reset time.
- Alternatively open the Codex usage dashboard in your account.
Tool
Codex and Claude give you a reset time and a usage figure, but not whether the two add up. Enter both: the calculator shows when your budget runs dry at that pace and puts the reset into your calendar.
Free and without an account. Your entries stay in this browser, and nothing here connects to your OpenAI or Anthropic account.
Two values are enough: the reset time from your account display and your current usage in percent.
Copy it exactly as /status prints it. Do not estimate.
The percentage from the same display. If it shows what is left, use 100 minus that number.
Optional. Enter the expiry date shown in your account or offer.
—
Enter a reset time above.
No values yet
Once the reset time and usage are in, this panel projects your pace forward.
The projection assumes steady usage. Real sessions are bursty: context size, tool calls and model choice all change the pace. Treat the number as orientation, not a promise.
Everything is stored in this browser only (localStorage). Clear removes all of it.
Work out which limit you hit first, otherwise you wait for the wrong reset.
The short-term budget. In Codex, local messages and cloud chats share this window.
You can tell by a reset that is only hours away.
The ceiling above it. It refills independently of the short-term window.
You can tell by a reset measured in days rather than hours.
Affects this one conversation, not your time budget.
You can tell because a new conversation fixes it immediately.
A separate billing path with its own rate and spending limits, not part of the subscription.
You can tell by error code 429 and a reference to an API key.
Alongside the regular windows, OpenAI occasionally resets limits for everyone, at milestones or after incidents. This list only collects events that can be documented publicly, each with its source.
"These resets are not on a weekly schedule. They usually happen around milestones like user growth or product anniversaries … Right now there is not a single official place to track upcoming resets." — OpenAI Support, 20 April 2026
Important: this list is incomplete. It counts only what is documented here, so the real gaps may be considerably shorter. No next reset can be derived from these numbers, and this page deliberately does not predict one.
Banked reset for every Codex and ChatGPT Work user at the milestone of 20 million active users.
20 Million+ Codex users: A free banked reset for everyone · OpenAI Developer Community
Hard reset for the milestone of 15 million users.
ChatGPT Business account did not receive the Codex 15M-user hard reset · OpenAI Developer Community
A second reset announced in the same breath for the following Monday.
Usage limits reset for all paid ChatGPT Work and Codex accounts to mark GPT-5.6 Sol.
Hard reset for the milestone of 8 million Codex users.
ChatGPT Business account did not receive the Codex 8M-user hard reset · OpenAI Developer Community
Not a reset but a limit change: the five-hour window was temporarily suspended for Plus, Pro and Business.
Clarification on the temporary removal of the Codex five-hour limit · OpenAI Developer Community
Introduction of flexible, bankable resets, in response to unsolicited resets hurting some users.
Flexible Rate Limit Resets for Codex and a method to get a Reset · OpenAI Developer Community
Reset for all accounts after a rolled-back optimisation had hurt cache hit rates in long sessions.
Codex Rate Limits reset incoming? (Beitrag 5) · OpenAI Developer Community
Reset announced after two causes of degraded GPT-5.5 quality in Codex had been fixed.
Codex Rate Limits reset incoming? · OpenAI Developer Community
Reset for all paid plans, announced as a celebration of a good week with GPT-5.5.
Codex Rate limits reset for all paid plans April 28, 2026 · OpenAI Developer Community
List last curated: 2026-08-24. New events are verified and added with a source.
Which limit you hit, and what waiting, credits or the API each cost you, is covered in detail in the article.
Read the article on Claude and Codex limits