[BUG] Sonnet only mode is at 23% usage but still I cannot use claude.
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
When using only the Sonnet model, the "All models" weekly aggregate limit reaches 100% and blocks all usage — even though the Sonnet-specific sub-limit shows only 23% used (77% remaining). This creates a ontradictory state where Sonnet is blocked despite having significant quota remaining in its own dedicated limit.
What Should Happen?
The Sonnet-specific sub-limit and the "All models" aggregate limit should work independently. If the Sonnet sub-limit is not exhausted, Sonnet should remain accessible regardless of the aggregate counter state. A user who exclusively uses Sonnet should never be blocked by an aggregate limit that their usage pattern could not have independently exhausted.
Error Messages/Logs
Steps to Reproduce
Steps to Reproduce
- Use Claude Code exclusively with the Sonnet model (no Opus or Haiku usage)
- Continue using Sonnet until the "All models" weekly aggregate limit reaches 100%
- Observe that the Sonnet-specific weekly sub-limit shows only 23% used (77% remaining)
- Attempt to continue using Sonnet
- Observe that usage is blocked despite the Sonnet sub-limit not being exhausted
Result: All Sonnet requests are blocked due to the aggregate limit being at 100%, even though the Sonnet-specific limit has 77% remaining.
Expected: Sonnet should remain usable since its dedicated sub-limit is not exhausted. The aggregate cap should not override a model-specific cap when only that model has been used.
Claude Model
Sonnet (default)
Is this a regression?
Yes, this worked in a previous version
Last Working Version
_No response_
Claude Code Version
2.1.142 (Claude Code)
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Terminal.app (macOS)
Additional Information
_No response_
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗