New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
HealthBench benchmark
AI model leaderboard for the HealthBench benchmark. Compare how large language models score on HealthBench, see the full ranking, and understand what this AI benchmark measures. GPT-5 currently leads with 67.2%. Realistic patient/clinician dialogues for health-domain LLM evaluation.
Leaderboard
| # | Model | Organization | Score | Variant | Source |
|---|---|---|---|---|---|
| #1 | GPT-5 | OpenAI | 67.2% | thinking-high-cited-ring-1t | official ↗ |
| #2 | Kimi K2 Thinking | Moonshot | 58% | thinking-no-tools | official ↗ |
| #3 | Ring-1T | Ant Group | 57.93% | thinking | official ↗ |
| #4 | gpt-oss-120b | OpenAI | 57.6% | high | official ↗ |
| #5 | gpt-oss-120b | OpenAI | 57.6% | high reasoning | official ↗ |
| #6 | Qwen3-235B-A22B-Thinking (Jul 2025) | Qwen | 55.56% | thinking-cited-ring-1t | official ↗ |
| #7 | DeepSeek-V3.1-Terminus | DeepSeek | 50.19% | thinking-cited-ring-1t | official ↗ |
| #8 | gemini-2.5-pro | Google DeepMind | 49.39% | cited-ring-1t | official ↗ |
| #9 | Kimi K2 | Moonshot | 43.8% | k2-0905-no-tools-cited-k2-thinking | official ↗ |
| #10 | gpt-oss-20b | OpenAI | 42.5% | high reasoning | official ↗ |
| #11 | gpt-oss-20b | OpenAI | 42.5% | high | official ↗ |
Frequently asked questions about HealthBench
What is the HealthBench benchmark?
Realistic patient/clinician dialogues for health-domain LLM evaluation.
How is the HealthBench benchmark scored?
HealthBench is scored using the score (%) metric, where a higher score is better. The maximum achievable score is 100.000. GenAIList aggregates reported scores from model providers and papers into a single ranked leaderboard.
Which AI model scores highest on HealthBench?
As of the latest reported scores on GenAIList, GPT-5 achieves the highest result on HealthBench with a score of 67.2%.
Is a higher HealthBench score better?
Yes. On HealthBench a higher score indicates better performance, so models near the top of the leaderboard are the strongest.