Claude Fable 5 Outperforms Competitors via High Biomedical Refusal Rates
July 14, 2026
Claude Fable 5 achieves top accuracy across eight biomedical benchmarks, including multimodal tasks, when excluding refusals. However, the model exhibits refusal rates ranging from 8.0% to 99.4% depending on the benchmark, a pattern not observed in GPT-5 or previous Claude iterations.
HOW THIS AFFECTS YOU
●
researcherYou must account for refusal rates when comparing model accuracy on sensitive benchmarks.
●
healthBe aware that high refusal rates may limit model utility in clinical decision support workflows.