[BUG] Don't use A/B tests with paying users, you will quickly use them

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?

Claude's performance has completely degraded in the past month, one thing I note is that it lies, tells something but does something else, and hallucinates a lot. I don't know if this is an A/B test, if it is, it's a pretty bad idea to do A/B tests on paying customers in real time. If it's not, then I need to find a different AI.

What Should Happen?

Claude's performance has completely degraded in the past month, one thing I note is that it lies, tells something but does something else, and hallucinates a lot. I don't know if this is an A/B test, if it is, it's a pretty bad idea to do A/B tests on paying customers in real time. If it's not, then I need to find a different AI.

Error Messages/Logs

Claude's performance has completely degraded in the past month, one thing I note is that it lies, tells something but does something else, and hallucinates a lot.  I don't know if this is an A/B test, if it is, it's a pretty bad idea to do A/B tests on paying customers in real time. If it's not, then I need to find a different AI.

Steps to Reproduce

Claude's performance has completely degraded in the past month, one thing I note is that it lies, tells something but does something else, and hallucinates a lot. I don't know if this is an A/B test, if it is, it's a pretty bad idea to do A/B tests on paying customers in real time. If it's not, then I need to find a different AI.

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.19 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_