Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Clinical Task Execution on MedAgentBench v2 (test)

76.9Accuracy

GRASP

11.27628.31345.3562.387May 28, 2026
Updated 5d ago

Evaluation Results

MethodLinks
2026.05
76.9
2026.05
74
2026.05
73.4
2026.05
72.9
2026.05
70.5
2026.05
70.3
2026.05
69.3
2026.05
67.8
2026.05
67.7
2026.05
67.2
2026.05
66.9
2026.05
65.1
2026.05
64.1
2026.05
63.5
2026.05
63.3
2026.05
62.5
2026.05
61.1
2026.05
60.9
2026.05
60.4
2026.05
59.9
2026.05
59.4
2026.05
57.8
2026.05
55.7
2026.05
54.2
2026.05
53.8
2026.05
51.6
2026.05
50.5
2026.05
44.6
2026.05
30.9
2026.05
25.3
2026.05
22.8
2026.05
20.4
2026.05
16.6
2026.05
14.7
2026.05
13.8