[BUG] Fix your f'ing quota/token counting algo

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?

1M model rollout: back to bugged f'ing usage again. The last 5 SMALL Python tool calls in my latest session had my usage go 86%->93%->97%->99%->locked out. This f'ing SERVER SIDE BUG is BACK AGAIN with this GOD DAMN skewed quota usage - fix your god damn app, I cannot even write /feedback about it, that is broken too.

Pathetic - this BS is court case worthy, I am seeking council.

What Should Happen?

I am pretty sure quota usage on tiny tool calls should not jump in skewed 4-7% chunks on Max 20x (your highest quota plan) - what do you think?

Error Messages/Logs

None - exactly the f'ing problem. Nobody knows how you are counting token usage/quota in the background - it's illegal.

Steps to Reproduce

1. Use Claude Code with Opus 4.6 1M model.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

2.1.74

Claude Code Version

2.1.75

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

_No response_