MedAgentBench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MedAgentBench OOD v2
87.1Accuracy
35
MedAgentBench v2 (test)
76.9Accuracy
35
MedAgentBench v2 (val)
77Accuracy
35
MedAgentBench OOD
80.6Accuracy
35
MedAgentBench (test)
88.8Accuracy
35
MedAgentBench (val)
86.2Accuracy
35
MedAgentBench
79.3Success Rate
24
MedAgentBench
87Accuracy
10
MedAgentBench
48.6Accuracy
7
MedAgentBench
73.3Accuracy
4
MedAgentBench
100Top-1 Acc
4
MedAgentBench
99Top-1 Accuracy
4
MedAgentBench (Adversarial)
0Unsafe Rate (Hong et al.)
3
MedAgentBench (Original)
0Unsafe Rate
3