Diagnostic Reasoning Generalization on CheXbench (test)
68.7Visual QA: Rad-Restruct (Top-1 Acc)MedRAX
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MedRAXModel Category=Multi-Agent, Venue=ICML’252026.04 | 68.7 | 82.9 | 52.6 | 68.1 | |
| XrayClawModel Category=Multi-Agent2026.04 | 66.3 | 85.6 | 62.1 | 70.7 | |
| CheXagentModel Category=Multi-Agent, Venue=AAAI’24 Sym’2026.04 | 57.1 | 78.1 | 59 | 64.7 | |
| GPT-4oModel Category=MLLMs-Based, Venue=Commercial2026.04 | 53.9 | 85.4 | 51.1 | 63.5 | |
| LLaVA-MedModel Category=MLLMs-Based, Venue=NeurIPS’232026.04 | 34.9 | 55.5 | 45.8 | 45.4 |