[BUG] Rate limit reached on ALL surfaces (Code + claude.ai + Deep Research) — Max 20x, 18% usage, non-peak
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?
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report
- [x] I am using the latest version of Claude Code
What's Wrong?
Max 20x subscriber ($200/month). Getting "API Error: Rate limit reached" on every Claude surface simultaneously:
- Claude Code CLI — rate limit on first prompt
- Claude.ai web chat — same error
- Deep Research — hits rate limit at ~50 turns, gets kicked out mid-research making it impossible to complete any research task
Usage dashboard at time of error:
- Session: 18% used (4hr 31min until reset)
- Weekly All Models: 45% used (resets Fri 2:00 PM)
- Sonnet only: 1% used
This occurs at all times of day, including off-peak hours (outside 5–11am PT).
Additional details:
- Rate limiting triggers consistently during peak hours (3PM–9PM Moscow time / 12PM–6PM UTC), but also occurs outside peak hours
- Even trivial requests (single short prompt, no file context, no tools) get rate limited immediately
- Deep Research becomes unusable — gets kicked out mid-research with rate limit error
- 2–3 days ago I was running heavy workloads (multiple agents, large codebases, MCP servers) with zero rate limit issues. Now even a simple "hello" triggers the error
- This is not a usage pattern issue — the dramatic change from "heavy usage works fine" to "zero usage gets blocked" within 24 hours points to a server-side state corruption
- Anthropic's own AI support agent confirmed this needs human investigation and that session-based limits cannot be reset by support team
I am paying $200/month and currently unable to use the product at all. This is not a matter of optimization or reducing token consumption — the service is fundamentally broken for my account.
Timeline
- 2–3 days ago: worked perfectly, heavy usage with multiple agents and MCP servers, no issues
- Now: rate limited on every request across all surfaces, including trivial prompts
What I've Tried
claude logout→claude login— no effect- Updated Claude Code to latest version
- Waited 12+ hours with zero usage — still rate limited
- Tried at various times of day (peak and off-peak)
- Contacted Anthropic support — AI agent confirmed need for human investigation
Why This Is a Server-Side Bug
The fact that rate limiting occurs on claude.ai web chat AND Deep Research (not just Claude Code) proves this is not a client configuration issue, MCP server problem, or context size issue. This is an account-level stuck rate limit state on the server side.
The sudden transition from "heavy usage with zero issues" to "completely blocked on trivial requests" within 24 hours is consistent with server-side rate limiter state corruption, not legitimate quota exhaustion.
Related Issues
- #29579 — Rate limit at 16% usage, Max plan,
rateLimitTier: null - #39581 — Rate limit at 0% usage, stuck for 7+ hours
- #39649 — Multiple anomalies since March 23, 2026
Environment
- Plan: Max 20x ($200/month)
- Claude Code version: Claude Code 2.1.87
- OS: Windows
Screenshots
What Should Happen?
All Claude surfaces (Claude Code, claude.ai, Deep Research) should work normally when usage dashboard shows 18% session and 45% weekly utilization. A Max 20x subscriber paying $200/month should be able to use the service without rate limit errors when well within their allocated quota
Error Messages/Logs
Steps to Reproduce
- Have an active Max 20x subscription ($200/month)
- Confirm usage dashboard shows low utilization (e.g. 18% session, 45% weekly)
- Open Claude Code CLI
- Type any trivial prompt, e.g.:
hello - Observe: "API Error: Rate limit reached"
- Open claude.ai in browser, send any message → same error
- Start a Deep Research task → gets rate limited and kicked out mid-research
- Run
claude logoutthenclaude login→ error persists - Wait 12+ hours with zero usage → error still persists
Note: This is an account-specific server-side issue and may not reproduce on other accounts. The reproduction value is that it demonstrates the error occurs across ALL Claude surfaces (not just CLI), which rules out any client-side cause.
Claude Model
Opus
Is this a regression?
Yes, this worked in a previous version
Last Working Version
_No response_
Claude Code Version
Claude Code 2.1.87
Platform
Anthropic API
Operating System
Windows
Terminal/Shell
Terminal.app (macOS)
Additional Information
_No response_
This issue has 9 comments on GitHub. Read the full discussion on GitHub ↗