Leaderboard
An open, live leaderboard for healthcare AI agents, supporting continuous evaluation and rolling updates
| Rank | Agent | ESL-Bench | MedHall-Bench | MedHarm-Bench | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Lookup | Trend | Comparison | Anomaly | Explanation | Factual | Contextual | Citation | Numerical | Relational | Implicit Dx | 2nd Opinion | Data Interp. | Coercion | Authority | |||||
 | gpt-5.4 | 64.5 | 50.4 | 74.1 | 62.4 | 76.7 | 58.9 | 71.2 | 90.1 | 77.1 | 27.3 | 66.8 | 76.3 | — | — | — | — | — | — |
 | theta-smart-expert | 59.6 | 57.0 | 65.7 | 62.9 | 42.3 | 69.8 | 73.1 | 88.6 | 56.9 | 59.6 | 69.1 | 81.1 | — | — | — | — | — | — |
 | theta-smart-miroflow | — | — | — | — | — | — | 75.2 | 89.6 | 65.0 | 35.9 | 77.4 | 82.8 | — | — | — | — | — | — |
4 | gpt-5.5 | 79.4 | 65.1 | 73.4 | 88.4 | 92.1 | 78.1 | — | — | — | — | — | — | — | — | — | — | — | — |
5 | glm-5.1 | 64.5 | 51.0 | 68.0 | 70.9 | 78.1 | 54.3 | — | — | — | — | — | — | — | — | — | — | — | — |
6 | gemini-3.5-flash | 60.2 | 39.9 | 69.4 | 64.8 | 71.4 | 55.4 | — | — | — | — | — | — | — | — | — | — | — | — |
7 | gemini4_1 | — | — | — | — | — | — | 70.0 | 80.7 | 50.6 | 44.2 | 71.4 | 80.8 | 78.4 | 80.0 | 85.3 | 64.7 | 84.0 | 90.0 |
8 | claude-sonnet-4.6 | — | — | — | — | — | — | 69.7 | 67.8 | 69.8 | 60.5 | 66.8 | 76.6 | — | — | — | — | — | — |
9 | gemini-3-pro-preview | — | — | — | — | — | — | 48.1 | 47.3 | 59.1 | 36.6 | 54.2 | 40.4 | — | — | — | — | — | — |
10 | deepseek-v4-pro | 59.5 | 41.1 | 79.0 | 59.1 | 67.7 | 50.6 | — | — | — | — | — | — | — | — | — | — | — | — |