[BUG] Claude misreports (lies about) token usage by selective ommission ("white lie")

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?

Most first time users of Claude quickly fall under a misimpression of its token usage, because it only shows output tokens.

What Should Happen?

Input tokens should also be reported ie.

25132 + 234 tokens in, 123 tokens out

Error Messages/Logs

Steps to Reproduce

Open claude
run anything

Claude Model

Not sure / Multiple models

Is this a regression?

No, this never worked

Last Working Version

_No response_

Claude Code Version

HEAD

Platform

Anthropic API

Operating System

Ubuntu/Debian Linux

Terminal/Shell

Other

Additional Information

_No response_