Skip to content

HealthBench length-adjusted score

HealthBench score after applying a verbosity penalty to model responses.

7 models

Higher is better.

  1. 1Claude Sonnet 5.5Anthropic65.4
  2. 2Claude Opus 5.5Anthropic60.6
  3. 3GPT-6.1 SolOpenAI58.5
  4. 4GPT-6 AstraOpenAI58.3
  5. 5Claude Opus 5Anthropic57.8
  6. 6GPT-6 LunaOpenAI54.5
  7. 7GPT-6 SolOpenAI53.2
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Length-adjusted 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.