[BUG] "API Error: Rate limit reached" is actually a stale auth issue - persists for hours past reset, fixed instantly by re-login

Resolved 💬 3 comments Opened Mar 26, 2026 by bl4dee Closed Apr 29, 2026

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?

After hitting a legitimate rate limit in a long Opus session, every subsequent interactive Opus request returned API Error: Rate limit reached, including brand new sessions. This persisted for many hours past every stated reset time. /status showed 29% weekly usage.

The fix was claude auth logout && claude auth login. Opus worked immediately after.
This was not a rate limit. It was a stale auth state being reported as a rate limit.

What Should Happen?

If the problem is auth, the error should say so. "Rate limit reached" when the fix is re-authenticating is misleading. At minimum, the error output should suggest running claude auth logout && claude auth login as a troubleshooting step when the rate limit state does not match /status usage.

Error Messages/Logs

Raw log entries from ~/.claude/projects/-home-blink/*.jsonl. 

All have "error":"rate_limit","isApiErrorMessage":true. Timestamps UTC.

Original session (d320e62e):
2026-03-25T07:02:05Z  "You're out of extra usage · resets 3am (America/Los_Angeles)"
2026-03-25T18:01:02Z  "You're out of extra usage · resets 3pm (America/Los_Angeles)"
2026-03-25T20:34:57Z  "You're out of extra usage · resets 3pm (America/Los_Angeles)"
2026-03-25T21:01:26Z  "You're out of extra usage · resets 3pm (America/Los_Angeles)"
2026-03-25T22:29:38Z  "You've hit your limit · resets 6pm (America/Los_Angeles)"
2026-03-26T03:33:19Z  "API Error: Rate limit reached"
2026-03-26T03:57:29Z  "API Error: Rate limit reached"
2026-03-26T04:11:10Z  "API Error: Rate limit reached"

Brand new sessions (fresh launches, not resumes):
2026-03-26T03:10:37Z  session 19073e48  "API Error: Rate limit reached"
2026-03-26T03:16:09Z  session 10ee9849  "API Error: Rate limit reached"
2026-03-26T05:11:15Z  session 3b8b940b  "API Error: Rate limit reached"

/status during this period:
Current session:  9% used
Current week:    29% used

After running claude auth logout && claude auth login, interactive Opus worked on the next attempt.

Steps to Reproduce

Run a long interactive Opus 4.6 session on Pro plan

Hit a legitimate rate limit

Wait past the stated reset time

All interactive Opus requests fail with "Rate limit reached", including new sessions

/status shows low usage

Run claude auth logout && claude auth login

Everything works immediately

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.76

Platform

Anthropic API

Operating System

Other Linux

Terminal/Shell

Other

Additional Information

The original session was a multi-hour reverse engineering project with heavy tool use (7MB jsonl, 15+ subagent files). It legitimately used a lot of tokens. The initial rate limit was expected. What was not expected was hours of false "rate limit" errors that only cleared with a re-login.

Related issues with the same pattern (low usage, persistent rate limit, account-specific):

#29579: rate limit at 16% usage on Max plan
#33094: rate limit on Max plan after updating to Opus 4.6
#33120: rate limit on Cowork only, resolved by switching accounts

If any of those users tried claude auth logout && claude auth login, I'd bet it would fix theirs too.

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗