Clinical Task Execution on MedAgentBench v2 (test)
76.9AccuracyGRASP
Evaluation Results
| Method | Links | |
|---|---|---|
| GRASPModel=gpt-oss-120b2026.05 | 76.9 | |
| SkillXModel=GPT-4.12026.05 | 74 | |
| Batch MemoryModel=GPT-4.12026.05 | 73.4 | |
| ExpeLModel=GPT-4.12026.05 | 72.9 | |
| No skillsModel=GPT-4.12026.05 | 70.5 | |
| GRASPModel=GPT-4.12026.05 | 70.3 | |
| Evo-MedAgentModel=GPT-4.12026.05 | 69.3 | |
| ExpeLModel=gpt-oss-120b2026.05 | 67.8 | |
| Seq. MemoryModel=GPT-4.12026.05 | 67.7 | |
| GRASPModel=Gemini 3.1 Flash Lite2026.05 | 67.2 | |
| Evo-MedAgentModel=gpt-oss-120b2026.05 | 66.9 | |
| SkillXModel=Gemini 3.1 Flash Lite2026.05 | 65.1 | |
| SkillXModel=gpt-oss-120b2026.05 | 64.1 | |
| Batch MemoryModel=GPT-5.4 (low)2026.05 | 63.5 | |
| GRASPModel=GPT-5.4 (low)2026.05 | 63.3 | |
| Evo-MedAgentModel=GPT-5.4 (low)2026.05 | 62.5 | |
| No skillsModel=gpt-oss-120b2026.05 | 61.1 | |
| Seq. MemoryModel=GPT-5.4 (low)2026.05 | 60.9 | |
| ExpeLModel=GPT-5.4 (low)2026.05 | 60.4 | |
| No skillsModel=GPT-5.4 (low)2026.05 | 59.9 | |
| Batch MemoryModel=gpt-oss-120b2026.05 | 59.4 | |
| SkillXModel=GPT-5.4 (low)2026.05 | 57.8 | |
| ExpeLModel=Gemini 3.1 Flash Lite2026.05 | 55.7 | |
| Evo-MedAgentModel=Gemini 3.1 Flash Lite2026.05 | 54.2 | |
| Seq. MemoryModel=gpt-oss-120b2026.05 | 53.8 | |
| Batch MemoryModel=Gemini 3.1 Flash Lite2026.05 | 51.6 | |
| Seq. MemoryModel=Gemini 3.1 Flash Lite2026.05 | 50.5 | |
| No skillsModel=Gemini 3.1 Flash Lite2026.05 | 44.6 | |
| GRASPModel=DeepSeek V4 Flash2026.05 | 30.9 | |
| Evo-MedAgentModel=DeepSeek V4 Flash2026.05 | 25.3 | |
| ExpeLModel=DeepSeek V4 Flash2026.05 | 22.8 | |
| No skillsModel=DeepSeek V4 Flash2026.05 | 20.4 | |
| Batch MemoryModel=DeepSeek V4 Flash2026.05 | 16.6 | |
| SkillXModel=DeepSeek V4 Flash2026.05 | 14.7 | |
| Seq. MemoryModel=DeepSeek V4 Flash2026.05 | 13.8 |