HealthBench Professional

OpenAI's HealthBench Professional: 525 clinician conversations covering consult, writing and research tasks, graded against physician-written rubrics.

Models: 15 current, 19 in total
Median: 63.0%
Range: 51.7%–68.1%

Score distribution

Lowest: 51.7%
Median: 63.0%
Highest: 68.1%

Leaderboard

Include deprecated models

Data from Artificial Analysis

Rank	Compare	Model	Score
1		Claude Fable 5.2 (Xhigh, Default Fallback) Anthropic	68.1%
2		Claude Fable 5.2 (Max, Default Fallback) Anthropic	67.7%
3		Claude Fable 5.2 (High, Default Fallback) Anthropic	67.6%
4		Claude Fable 5.2 (Medium, Default Fallback) Anthropic	67.0%
5		Claude Fable 5.2 (Low, Default Fallback) Anthropic	66.9%
6		GPT-6 Astra (Max) OpenAI	66.0%
7		GPT-6.1 Sol (Max) OpenAI	65.1%
8		Claude Sonnet 5.5 (Max, Default Fallback) Anthropic	63.0%
9		GPT-6 Luna (Max) OpenAI	61.9%
10		Claude Opus 5.5 (Max, Default Fallback) Anthropic	59.9%
11		Claude Haiku 5.5 (Max) Anthropic	58.4%
12		Grok 4.7 (Xhigh) SpaceXAI	55.8%
13		Gemini 4 Argon (High) Google	55.1%
14		GLM-5.3 (Max) Z AI	52.8%
15		Gemini 3.8 Flash (High) Google	51.7%

1
Claude Fable 5.2 (Xhigh, Default Fallback)
Anthropic
68.1%
2
Claude Fable 5.2 (Max, Default Fallback)