A model upgrade is not a drop-in for your agent loop

A Claude ".1" is usually a model-id swap. Fable 5.1 is the release that breaks that habit in agent loops.

Forced tool_choice paths return 400. Thinking blocks do not survive cross-model routing the way teams assume. Editing earlier turns can invalidate reasoning state.

If your pipeline still forces tools, replays history across models, or mutates prior messages, run the 30-minute audit before you move production traffic.

Friend link: https://levelup.gitconnected.com/claude-fable-5-1-every-breaking-change-in-your-agent-loop-9be40b5fe787?sk=89d27a4f27b2c93ddbb9e4e0f3b696fe

Canonical: https://medium.com/gitconnected/claude-fable-5-1-every-breaking-change-in-your-agent-loop-9be40b5fe787