Benchmark
MedAgentBench
Evaluate agents completing clinical tasks in a virtual electronic-health-record environment.
Published leaderboard
Explore scores and evaluation settings on the publisher’s leaderboard.
Open leaderboardCoverage
The original suite has 300 physician-written tasks across 10 categories and an interactive FHIR environment with 100 de-identified patient profiles.
Methodology
Agents query records and execute actions in the virtual EHR. The official project publishes overall success rates plus separate query and action success rates.