CODEX · CLAUDE CODE
When do Codex and Claude Code limits reset?
Find your own reset time, understand a public reset report, and plan what to do while you wait.
The short answer
Your account’s usage screen is the place to check your own reset time. TokenReset tracks public Codex and Claude Code reports and offers a local weekly planner; its historical estimates cannot confirm when your account will recover.
Three different meanings of “reset”
| What you see | What it means | What to check |
|---|---|---|
| Account reset time | A usage window shown for your account and plan. | Your usage screen or limit message. |
| Public reset announcement | A source reports an event for a stated audience. | The original post, eligible plans and publication time. |
| Historical estimate | A calculation from gaps between earlier public reports. | Sample size, source coverage and the method below. |
Start with your own account
- Codex: open the usage dashboard. In Codex CLI, /status shows your remaining limits. Codex ↗
- Claude Code: read the limit message for the reset time. /usage helps you inspect usage. Claude Code ↗
Which limit am I waiting for?
Codex limits depend on your plan. Read each window shown in your dashboard; a public report cannot confirm your balance.
Claude Code separates session cost from plan usage in /usage. Clearing a conversation resets session statistics; read the limit message for your allowance’s reset time.
Why a public reset countdown can differ
A public announcement describes an event and its stated audience. It cannot see your account. Check the original post, eligible plan and time zone before treating it as relevant to you. A historical estimate is not a scheduled reset.
Browse public reset reports →Make the wait useful
Write down the next task, collect the files it needs and reduce its scope. If your account shows a weekly reset, enter that time and your used percentage in the planner. Nothing connects to your account, and your inputs stay in this browser.
Open the free weekly planner →Related reset tools and explanations
How the Codex public-report estimate is calculated
We use up to 180 days of recorded usage-reset and reset-credit reports, count each original post once and calculate positive gaps between posts. If the current wait is t days, only past gaps longer than t are comparable.
Next-24-hour historical frequency = comparable gaps ending within t + 1 days ÷ all gaps longer than t.
Illustration only: if 40 past gaps exceed the current wait and 10 ended within the following day, the frequency is 10 ÷ 40 = 25%. These are example numbers, not today’s result.
The displayed window spans the 25th–75th percentiles of comparable gaps. We withhold the Codex estimate when there are fewer than 12 total gaps or 10 comparable gaps, or the feed is stale or incomplete. These are past frequencies, not a calibrated probability, an official schedule or a personal quota forecast. Claude uses a separate model and can withhold its estimate when editorial coverage is old.
Sources and editorial responsibility
TokenReset is an independent project, not OpenAI or Anthropic. Public reports arrive through third-party feeds and link to original posts; a successful feed check is not proof of complete X coverage or an account reset. Check the original source before acting. The guide and calculation explanation are prepared by TokenReset.