Usage spikes drastically after /compact command and even after running claude

Resolved 💬 3 comments Opened Apr 24, 2026 by Jagadish-Rapuru Closed Apr 28, 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?

Environment

  • Claude Code version: 2.1.119
  • Plan: Claude Max 5x ($100/month)
  • Model: Opus 4.7 (1M context)
  • Reset time: 6:00 PM

Issue Summary

After running the /compact command with a small prompt, usage jumped
approximately 34% in a single operation. Before /compact the usage
was at a lower level and after it completed, usage showed 64%.

Shortly after, running "claude --version" which is a local command
that requires NO API call and NO server communication, caused usage
to jump from 64% to 100%.

What Should Happen?

  • /compact with a small prompt should have minimal usage impact
  • "claude --version" is a local command and should consume

ZERO tokens and ZERO usage under any circumstances

  • Users should receive a warning before /compact runs that it

may consume significant usage

Error Messages/Logs

No error messages shown. Usage bar visually jumped to 100% 
with a warning icon after running claude --version

Steps to Reproduce

  1. Start a Claude Code session with some conversation history
  2. Run /compact command with a small minimal prompt
  3. Observe usage bar jumping approximately 34% immediately
  4. Run "claude --version" in terminal
  5. Observe usage jumping from 64% to 100% despite no API call being made

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.119

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

This is directly affecting paid subscribers on the Claude Max plan
($100/month). Usage limits are being consumed by operations that
either should not count at all or are disproportionately high
without any warning. This results in lost paid usage with no
recourse for refund as per Anthropic support response.

This appears to be part of a wider pattern reported by multiple
Max plan subscribers where usage limits are being hit unexpectedly
faster than before.

View original on GitHub ↗

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