[BUG] Fable 5.1 burns tokens faster than 5.0 100% of limit in 20 mins.

### Preflight Checklist

- [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) 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?

Fable 5.1 medium burns through 100% of claude code limit in under 20 mins!!! its even less token efficient than 5.0 even though you report it as more efficient.

### What Should Happen?

Fable 5.1 medium burns through 100% of claude code limit in under 20 mins!!! its even less token efficient than 5.0 even though you report it as more efficient.

### Error Messages/Logs

```shell

```

### Steps to Reproduce

Fable 5.1 medium burns through 100% of claude code limit in under 20 mins!!! its even less token efficient than 5.0 even though you report it as more efficient.

### Claude Model

Other

### Is this a regression?

Yes, this worked in a previous version

### Last Working Version

_No response_

### Claude Code Version

2.1.257

### Platform

Anthropic API

### Operating System

Ubuntu/Debian Linux

### Terminal/Shell

VS Code integrated terminal

### Additional Information

_No response_

No source file, test, logs, or exact Fable model identifier is provided. Start by reproducing the report with Claude Code 2.1.257 in the VS Code terminal on Ubuntu/Debian, comparing Fable 5.1 medium with 5.0 while recording token usage. Done means the regression is reproducible with enough usage data to identify the affected component and verify a fix.

Written by the indexing model from the issue text.