Benchmark
AgentClinic
Evaluate tool-using medical agents through simulated patient encounters and multimodal clinical evidence.
Explore the original evidence
Explore the original evaluation, its methods, and reported findings.
Open evalCoverage
Published settings span nine medical specialties, seven languages, medical-record cases, and multimodal clinical cases.
Methodology
A doctor agent gathers information through patient dialogue, measurements, and tools before diagnosing. Evaluation varies doctor model, patient simulator, tools, and case source.