AnalysisAI ModelsJuly 3, 2026

Claude Fable 5 Isn't Nerfed. The Router Is Just Paranoid

BridgeBench's debugging score for Claude Fable 5 dropped from 86.2 to 25.9 after reinstatement, but the drop was due to the safety classifier routing tasks to Opus 4.8. Arena AI found performance flat, with some categories improving, and Anthropic acknowledged false positives will occur with no timeline for refinement.

1 source