Whether OpenAI Codex / ChatGPT Work and Anthropic Claude have publicly reset usage limits — staff-announced, provider-wide 'hard resets' — plus the time since the last public reset and a heuristic 48-hour forecast, each with a link to the source announcement.
As of 2026-07-21T07:16:04.314Z, Codex — last public reset 3.2d ago; Claude — last public reset 5.1d ago.
No. A confirmed public/global RESET means many paid users had usage replenished; it does not guarantee your individual account is full, and it is not your personal 5-hour or weekly rolling window.
The personal 5-hour limit is a rolling window that starts from your first message. RESET Radar does not track your personal window — it tracks provider-wide public 'bonus' resets, which are separate and irregular.
The weekly limit is a rolling 7-day window per account. Public resets tracked here are separate staff actions, not your weekly rollover.
Claude uses a rolling ~5-hour window plus a weekly cap per account. The public resets tracked here are separate, provider-wide events.
A banked reset is a redeemable reset offer announced by staff; it is not an automatic refill. RESET Radar shows it as a distinct 'banked' state, never a green public reset.
It is a deterministic heuristic (a Weibull renewal survival model) fitted to confirmed hard-reset intervals, shown with an uncertainty range and its sample size. It is not official and never a confirmation — see the Methodology page.
Across the tracked history, roughly every 12.2 days on average — but this varies widely, and resets tend to cluster around model launches.
No. It is an independent utility, not affiliated with OpenAI, Anthropic, xAI, Moonshot, z.ai, or Google. It only reads public announcements and never logs into your AI accounts.
Independent utility. Not affiliated with OpenAI, Anthropic, xAI, Moonshot, z.ai, or Google. A confirmed public reset does not guarantee your personal quota is full.