Leaderboard
An open, live leaderboard for healthcare AI agents, supporting continuous evaluation and rolling updates
| Rank | Agent | MetAgentBench | |||||
|---|---|---|---|---|---|---|---|
| Data Suff. | Risk | Differential | Safe Action | Closed Loop | |||
 | gpt-5.6-terra | 56.9 | 51.6 | 69.8 | 43.9 | 52.5 | 66.7 |
 | gpt-5.6-terra-high | 55.5 | 51.3 | 63.3 | 41.7 | 52.0 | 69.1 |
 | deepseek-v4-flash-xhigh | 55.5 | 54.8 | 64.0 | 44.4 | 44.7 | 69.4 |
4 | deepseek-v4-pro | 53.6 | 54.7 | 60.5 | 45.7 | 45.7 | 61.3 |
5 | kimi-k3-high | 53.5 | 53.0 | 62.8 | 45.6 | 48.0 | 58.0 |
6 | glm-5.3-high | 51.6 | 49.4 | 62.2 | 43.0 | 44.5 | 58.9 |
7 | kimi-k3 | 49.2 | 49.8 | 56.7 | 43.7 | 40.3 | 55.7 |
8 | minimax-m3 | 37.5 | 38.7 | 34.0 | 34.9 | 34.8 | 45.3 |