List your company

Open-ended Medical Consultation on HealthBench Hard

46.2Total ScoreGPT-5 (thinking)
-1.84810.62623.135.574Oct 17, 2025

Evaluation results

MethodLinks
GPT-5 (thinking)Model Category=Proprietary Models, Model Type=Thinking2025.1046.2————————————
InfiMed-ORBIT-4B (28k)Model Category=Open-source Models (< 10B), Training Sample Size=28k2025.1037.351.650.541.910.830.943.813.438.148.942.131.349.1
Baichuan-M2-32BModel Category=Open-source Models (> 10B)2025.1034.545.639.535.621.33240.919.941.344.651.619.348
InfiMed-ORBIT-4B (8k)Model Category=Open-source Models (< 10B), Training Sample Size=8k2025.1033.644.649.134.69.32842.610.433.64246.732.350.5
InfiMed-ORBIT-4B (2k)Model Category=Open-source Models (< 10B), Training Sample Size=2k2025.1027.539.937.830.26.226.632.26.631.838.145.316.843.7
Qwen3-30B-A3B-ThinkingModel Category=Open-source Models (> 10B), Model Type=Thinking2025.1016.121.420.4158.916.720.46.533.711.653045.5
GPT-4.1Model Category=Proprietary Models2025.1013.220.512.312.19.714.912.317.530.5070.6060.5
Qwen3-30B-InstructModel Category=Open-source Models (> 10B), Model Type=Instruct2025.1013.118.312.914.717.919.49.528.528.5045.2033.7
Qwen3-4B-Instruct (base)Model Category=Open-source Models (< 10B), Model Type=Instruct (base)2025.1079.38.57.108.612.25.124.10.857.5045
Qwen3-4B-ThinkingModel Category=Open-source Models (< 10B), Model Type=Thinking2025.105.214.412.52.403.58.5023.2042.5039.6
Qwen-2.5-7B-InstructModel Category=Open-source Models (< 10B), Model Type=Instruct2025.10000000006.4045.2033.7