Last updated: 2026-09-14
❯ claude✗ Claude usage limit reached. Your limit will reset at 3:00 PM.❯ /costSession: 2.1M tokens · $0.00 (subscription)5-hour window ▓▓▓▓▓▓▓▓▓▓ 100% · resets 15:00weekly window ▓▓▓▓▓▓▓░░░ 71% · resets Sat 00:30# shared across claude.ai · Claude Code · Desktop
Two limits: a rolling 5-hour window and a fixed weekly window
Every paid Claude plan meters usage two ways at once. The 5-hour window is a rolling session limit: it starts with your first message and caps how much you can send over the next five hours. The weekly window is a separate, larger cap over a seven-day period.
You can hit either one. Running into the 5-hour cap pauses you for a few hours; running into the weekly cap pauses you until the week resets, and a five-hour wait will not bring it back.
It is one shared pool across claude.ai, Claude Code and Desktop
The single most surprising thing: usage is not counted per app. Messages in the claude.ai web app, in Claude Code, and in the Claude desktop app all draw from the same limit. A heavy Claude Code session in the morning can leave you rate-limited in the web app in the afternoon.
Model choice matters too — Opus-class models consume the pool far faster than smaller ones, which is why Max users on the biggest model can hit limits that Pro users rarely see.
When does it reset?
The 5-hour window resets five hours after the first message that opened it — it rolls, so it is always “five hours from when you started”, not a fixed clock time.
The weekly window resets on a fixed schedule tied to your account, not rolling. The exact day and time are shown in Settings → Usage on claude.ai (for example “Sat 12:30 AM”); it is account-specific, so there is no single universal reset time. If you hit the weekly cap, that fixed reset is the only thing that restores it.
Pro, Max 5× and Max 20×
The plans differ mainly in how large those windows are. Pro ($20/mo) is sized for everyday chat and light coding. Max at 5× and 20× multiply the caps for heavy Claude Code use. Because everything shares one pool, the practical question is not “which app” but “how much total, on which model” — the /cost command in Claude Code shows what the current session has spent.
See your own usage right now
The numbers that matter are your own, and you do not have to open claude.ai to read them. Export your claude.ai session cookie and run it through the checker below — it reports your plan and exactly how much of the 5-hour and weekly windows you have used, and when each resets. That is the fastest way to tell whether you are actually rate-limited or just hit a transient error.
Claude usage limits — questions
Almost always because the weekly pool is shared across claude.ai, Claude Code and Desktop, and a large-model (Opus-class) Claude Code session burns it quickly. Check what is spending it with /cost in Claude Code, and consider a smaller model for routine work.
Yes. Smaller/faster models consume the shared pool much more slowly than the largest model, so switching for routine edits meaningfully extends how far your 5-hour and weekly windows stretch.
No. The 5-hour and weekly windows apply to the Pro/Max subscription. The API is metered separately as pay-as-you-go by tokens — which is also why an ANTHROPIC_API_KEY set in your shell can quietly bill you instead of using your subscription.
Yes — paste your session cookie into the checker on this site and it reports your plan and both usage windows. It reads the same numbers claude.ai shows in Settings → Usage.
The 5-hour window rolls (five hours after your first message); the weekly window is fixed to a day and time on your account, shown in Settings → Usage. Hitting the weekly cap can only be cleared by that fixed weekly reset.
How much of your limit is left?
Paste your claude.ai session cookie and read your plan plus exactly how much of the 5-hour and weekly windows you have used, and when they reset.