Skip to content

HealthBench raw score

Raw score on realistic multi-turn healthcare conversations graded against expert-written rubrics.

7 models

Higher is better.

  1. 1Claude Sonnet 5.5Anthropic69.4
  2. 2Claude Opus 5.5Anthropic68.1
  3. 3Claude Opus 5Anthropic67.1
  4. 4GPT-6 AstraOpenAI56.9
  5. 5GPT-6.1 SolOpenAI56.7
  6. 6GPT-6 LunaOpenAI50.0
  7. 7GPT-6 SolOpenAI47.1
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Raw rubric score
Tasks
5,000 multi-turn patient conversations
Authors
OpenAI (2025)

Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.