Treatment Planning on HealthBench (treatment-related conversations)
48.94Overall ScoreTheraAgent
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| TheraAgentModel Category=Proprietary Models2026.05 | 48.94 | 47.49 | 55.63 | 44.65 | 55.29 | 44.8 | 51.72 | 37.16 | |
| Claude-4-SonnetModel Category=Proprietary Models2026.05 | 44.28 | 35.1 | 46.5 | 40.91 | 50.64 | 40.63 | 40.86 | 36.26 | |
| Gemeni-2.5-ProModel Category=Proprietary Models2026.05 | 43.49 | 34.42 | 44.48 | 38.85 | 51.46 | 41.32 | 39.49 | 34.08 | |
| DeepSeek-R1Model Category=Open-Source Models2026.05 | 42.94 | 39.53 | 48.85 | 39.02 | 48.16 | 41.89 | 47.29 | 31.97 | |
| Kimi-K2Model Category=Open-Source Models2026.05 | 42.71 | 33.38 | 41.79 | 38.73 | 52.5 | 41.56 | 41.1 | 33.29 | |
| Grok-3Model Category=Proprietary Models2026.05 | 42.51 | 34.16 | 41.97 | 37.83 | 54.97 | 41.07 | 38.84 | 37.62 | |
| Qwen3-235B-A22BModel Category=Open-Source Models2026.05 | 40.24 | 32.59 | 43.44 | 34.91 | 50.3 | 40.1 | 37.51 | 31.42 | |
| OpenAI-o4-miniModel Category=Proprietary Models2026.05 | 39.46 | 32.25 | 40.85 | 35.61 | 52.11 | 40.78 | 44.19 | 31.14 | |
| Baichuan-M2-32BModel Category=Medical Specialized Models2026.05 | 38.76 | 35.23 | 41.47 | 33.38 | 46.11 | 36.1 | 37.55 | 34.04 | |
| GPT-4.1Model Category=Proprietary Models2026.05 | 34.48 | 21.65 | 33.24 | 27.87 | 45.92 | 36.66 | 31.42 | 30.03 | |
| OpenAI-o1Model Category=Proprietary Models2026.05 | 32.03 | 21.14 | 33.77 | 26.85 | 40.25 | 34.95 | 29.89 | 30.92 | |
| MedCritical-7BModel Category=Medical Specialized Models2026.05 | 24.73 | 20.04 | 35.33 | 25.52 | 22.2 | 33.05 | 31.73 | 32.03 | |
| Llama3-Med42-70BModel Category=Medical Specialized Models2026.05 | 24.45 | 15.19 | 28.97 | 22.48 | 32.43 | 33.87 | 25.78 | 31.2 | |
| UltraMedical-70BModel Category=Medical Specialized Models2026.05 | 23.45 | 17.7 | 30.77 | 20.39 | 20.28 | 29.54 | 25.03 | 30.99 | |
| GPT-4oModel Category=Proprietary Models2026.05 | 19.35 | 12.49 | 20.22 | 19.57 | 24.8 | 25.92 | 29.13 | 26.68 |