Repository navigation
[BUG] Incorrect 5-h and weekly usage being recorded #40983
Description
Preflight Checklist
- I have searched existing issues and this hasn't been reported yet
- This is a single bug report (please file separate reports for different bugs)
- I am using the latest version of Claude Code
What's Wrong?
My weekly window just reset.
I used 75,721 tokens in a single session.
After 76k tokens~ my hourly window is 100% used and my weekly window shows 10% usage. This cant be correct.
whatever version was running on Thursday 26th March was fine. On Friday evening this issue started. I had to wait until Monday's weekly reset to check again, and the issue still persists.
What Should Happen?
I expect the 5h window to allow more usage than 76k tokens and for that not to be 10% of the weekly limit
Error Messages/Logs
Steps to Reproduce
- use claude code for 5min
- see 10% of your weekly limit disappear
Claude Model
Opus
Is this a regression?
Yes, this worked in a previous version
Last Working Version
whatever was latest last thursday.
Claude Code Version
2.1.87
Platform
Anthropic API
Operating System
Ubuntu/Debian Linux
Terminal/Shell
Other
Additional Information
No response
Reacted by rpawar-1412, Davi Samora, maksam07, Marcos Vinícius, napir-xu and Dorian Massoulier
Activity
github-actions commentedon Mar 30, 2026
Found 3 possible duplicate issues:
This issue will be automatically closed as a duplicate in 3 days.
🤖 Generated with Claude Code
maksam07 commentedon Mar 30, 2026
I waited three hours for my five-hour limit to be updated, and then in just 15 minutes, I had 95% used. And this is with a MAX subscription. What's going on? It's like I'm wasting money.
mono-ped commentedon Mar 30, 2026
Window reset, went to the web UI and wrote "test". Weekly usage jumped from 10% to 23% hourly window instantly filled
marlvinvu commentedon Apr 4, 2026
My Claude says your 75K tokens hitting 100% hourly has an explanation from #40584: the client-side rate limiter counts CONTEXT SIZE (not just your tokens) against the budget. With Opus 1M context, the context itself can consume most of the budget before you type anything. The false rate limiter issue was partially addressed in v2.1.91 (cache bugs fixed) but the client-side context counting remains — see the April 3 update on #38239 confirming 151 synthetic rate limit entries with 0 API calls still active on v2.1.91.
github-actions commentedon May 14, 2026
Closing for now — inactive for too long. Please open a new issue if this is still relevant.
github-actions commentedon Jul 3, 2026
This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.