Health AI Benchmarks

All boards

6 boards, each ranking models on its own results.

Answers graded against expert rubrics

Safety under pressure

Capability indexes

Retired evaluations

History only: no comparable results on current models, so no board.

  • OpenAI Dynamic Mental Health Evaluations: Removed from the index on 2026-09-08 at the owner's request: not treated as a real, established benchmark.
  • HealthBench Consensus: near-saturated physician-consensus baseline; frontier runs stopped reporting it separately
  • MedQA / MultiMedQA: exam-style multiple choice, saturated above 95 percent since 2025; archived by its trackers
  • AgentClinic: no public frontier-model results since 2025
  • CRAFT-MD: no public frontier-model results since 2025
  • MedAgentBench: v2 lives on inside the MAST composite; the standalone board has no current frontier rows
  • SDBench / MAI-DxO: Microsoft's 2025 sequential-diagnosis study was not re-run on current models
  • Open Medical-LLM Leaderboard (Hugging Face): built on saturated exam sets; no frontier submissions in 2026
  • MedArena: clinician preference arena; ratings pool too thin on current frontier models to quote
  • AMIE evaluations: Google DeepMind research prototypes, never opened to cross-vendor comparison
  • LiveClin, PrIME-LLM, MedMCP-Calc: single studies with two or fewer current-frontier rows; tracked for a future qualifying update
  • EHR-Complex: not run on current models; newest result is from February 2026 and no Claude 5, GPT-6 or Gemini 4 model has a score
  • MedHELM: not run on current models; the newest Claude on the board is Opus 4.6 and the task set adds little beyond our other clinical boards
  • MedPIC-Bench: a single study that has not been run on current models; newest Claude result is Sonnet 4.6
  • HealthAdminBench: the official board still lists only the five models from the April 2026 paper; later vendor runs used a different setup and are not comparable
  • HealthBench Hard: only OpenAI and Meta publish scores; no Anthropic or Google model has an official result, so the board cannot compare labs
  • WHBench: not run on current models; newest results are from March 2026