Medical
MedAgentBoard
Multi-agent collaboration vs. single-LLM and conventional methods on medical tasks.
1,606items
13subjects
53%observed
Modelsubject type
Response matrix
Every model, scored item by item.
Each row is an AI model and each column an item, ordered so the strongest models and easiest items gather toward one corner. 13 subjects × 1,606 items, 53% of cells evaluated.
Fit to width. Hover for subject & item; click a cell for details.
lowhighUnobserved
Scale: 0 to 1, by task: medical-QA accuracy (0/1), lay-summary ROUGE-L F1, or predicted in-hospital-mortality probability. Mixed metrics, each already on a 0–1 scale.
Sample items
What the questions look like — and how subjects answer.
A spread of items across the difficulty range. This benchmark does not publish per-answer traces, so each item shows which subjects succeeded.
Subjects
The models, agents, and reward models evaluated.
13 subjects, ranked by mean response across this benchmark's items.
- 1b3ba62130.834
- 2bf4b4ba90.798
- 3329114c30.783
- 4668ea5600.781
- 529bb30170.777
- 6e2c30d590.763
- 798a84b580.739
- 8a28d84370.738
- 9641850fe0.737
- 10db87218d0.588
- 11fbb7b5bf0.316
- 12d863de040.314
- 13e309757b0.307