Usage limit discrepancy: Claude Code blocked at "100% used" while claude.ai Settings shows "44% used" on same Pro account

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?

Bug Report Summary — Claude Code / claude.ai Usage Discrepancy
Title: Usage limit discrepancy: Claude Code blocked at "100% used" while claude.ai Settings shows "44% used" on same Pro account
Description:
On the same Claude Pro account, at the same time, two usage displays show significantly different values:

claude.ai Settings → Usage (Image 1): "Current session: 44% used, resets in 2 min"
Claude Code /usage (Image 2): "Current session: 100% used, resets 12:59pm (Europe/Sofia)"

Claude Code blocked further input citing the limit was reached, while the claude.ai interface indicated substantial remaining capacity (44% used, near-immediate reset).
This contradicts the documentation stating that usage across all Claude product surfaces (claude.ai, Claude Code, Claude Desktop) counts toward the same shared limit.
Additional context from /usage (Image 2):

Total session cost: $53.37
Total duration (API): 4h 5m 52s
Total duration (wall): 23h 17m 47s
Current week (all models): 31% used
100% of usage came from sessions active 8+ hours
22% of usage was at >150k context

The large gap between API duration (~4h) and wall duration (~23h) suggests the "current session" tracking in Claude Code may be counting an extended/long-running session window that doesn't align with the rolling 5-hour session reflected in claude.ai's usage panel.
Expected behavior: Usage percentages and reset times shown in Claude Code /usage and claude.ai Settings → Usage should be consistent for the same account/time, per documentation stating they share a single limit.
Actual behavior: Claude Code reports 100% used / blocked, claude.ai reports 44% used / near reset — a ~56 percentage point discrepancy on the same account at the same time.
Similar reported issues: #24335, #54750, #61828 — all describe the same class of Claude Code vs claude.ai usage panel mismatch, with Claude Code blocking while claude.ai shows significant headroom.
Attachments: Two screenshots (claude.ai Settings → Usage panel; Claude Code /usage output) showing the discrepancy side by side.

What Should Happen?

I am working on a project

Error Messages/Logs

Steps to Reproduce

Just use it and look

Claude Model

Sonnet (default)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.177 (Claude Code)

Platform

Other

Operating System

Ubuntu/Debian Linux

Terminal/Shell

VS Code integrated terminal

Additional Information

<img width="1125" height="850" alt="Image" src="https://github.com/user-attachments/assets/96bd4b53-02d4-4e81-8673-262ea09b9181" />

<img width="1173" height="863" alt="Image" src="https://github.com/user-attachments/assets/d1c725c0-dae4-46ba-8418-34b22b1db3e5" />