[BUG] Claude Code "Usage limit reached" with 0% on 5-hour window and 5% weekly usage

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?

<img width="1340" height="297" alt="Image" src="https://github.com/user-attachments/assets/5bc68b7d-db27-46ba-b8aa-89fe7d9d6750" />

<p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">Claude Code is blocking every prompt with <strong>"Usage limit reached"</strong> despite the usage panel reporting almost no consumption across any meter.</p>
<p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">At the moment of the error, the plan usage panel showed:</p>
<div class="overflow-x-auto w-full px-2 mb-6">
Meter | Usage | Reset
-- | -- | --
5-hour limit | 0% | —
Weekly · all models | 5% | 12h
Weekly · Claude Design | 0% | —
Sonnet only | 2% | 12h

</div>
<p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">I have used Claude Code only <strong>once this week</strong> before hitting this. The block is not consistent with the displayed usage or with the published limits for my plan. The usage percentages do not advance after each rejected prompt, suggesting no request is actually being processed — yet the rate-limit error persists across new attempts.</p>

What Should Happen?

With 0% on the 5-hour window and 2% on the Sonnet weekly meter, prompts should be processed normally. The rate limit should only trigger when a meter actually approaches 100%.
Actual Behavior
Every prompt is rejected with "Usage limit reached" while all displayed usage meters remain in single digits. The error appears decoupled from actual quota consumption.

Error Messages/Logs

Steps to Reproduce

Open Claude Code in a local git repository (main branch, worktree).
Confirm the plan usage panel shows low utilization (in my case <5% on every meter, 0% on the 5-hour window).
Submit any prompt with Sonnet 4.6 (Medium reasoning) selected.
Receive Usage limit reached and observe the request is blocked.
Submit additional prompts — error continues, usage meters do not move.

Claude Model

Not sure / Multiple models

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

latest

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

IMPACT;

A paid subscriber unable to make a single

<img width="1340" height="297" alt="Image" src="https://github.com/user-attachments/assets/14ba26e9-189c-401f-b697-0d9d47b6617d" />

request after one prior use in a week effectively has no working product. The discrepancy between displayed quota and enforced quota also makes plan-tier decisions impossible — there is no meaningful signal for when to upgrade.