Evidence quality, not model scale, cuts clinical LLM errors

Answer synthesized by Claude Sonnet 4.6 over articles curated by our newsroom. Each [N] links directly to the source.