ResearchBenchmarksDiagnostic Accuracy on PubMed VivaBenchFollow43.5AccuracyMeDxAgent22.28427.79233.338.808Jun 2, 2026Evaluation ResultsMethodMethodLinksAccuracyMeDxAgentModel=Gemini-2.5-ProModel=Gemini-2.5-Pro2026.0643.5MeDxAgentModel=GPT-4oModel=GPT-4o2026.0643.3MeDxAgentModel=o4-miniModel=o4-mini2026.0637.7VivaBenchModel=Gemini-2.5-ProModel=Gemini-2.5-Pro2026.0635VivaBenchModel=o4-miniModel=o4-mini2026.0632VivaBenchModel=GPT-4oModel=GPT-4o2026.0623.1