Clinical Task Execution on MedAgentBench v2 (val)
77AccuracyGRASP
Evaluation Results
| Method | Links | |
|---|---|---|
| GRASPModel=gpt-oss-120b2026.05 | 77 | |
| GRASPModel=GPT-4.12026.05 | 73.8 | |
| ExpeLModel=GPT-4.12026.05 | 72.5 | |
| SkillXModel=GPT-4.12026.05 | 72.1 | |
| GRASPModel=GPT-5.4 (low)2026.05 | 71.2 | |
| ExpeLModel=GPT-5.4 (low)2026.05 | 70.4 | |
| GRASPModel=Gemini 3.1 Flash Lite2026.05 | 68.8 | |
| Seq. MemoryModel=GPT-4.12026.05 | 68.8 | |
| Evo-MedAgentModel=GPT-5.4 (low)2026.05 | 68.8 | |
| SkillXModel=GPT-5.4 (low)2026.05 | 68.8 | |
| Batch MemoryModel=GPT-4.12026.05 | 68.3 | |
| SkillXModel=gpt-oss-120b2026.05 | 67.5 | |
| Evo-MedAgentModel=GPT-4.12026.05 | 66.7 | |
| No skillsModel=GPT-4.12026.05 | 65.9 | |
| ExpeLModel=gpt-oss-120b2026.05 | 65.5 | |
| SkillXModel=Gemini 3.1 Flash Lite2026.05 | 65.4 | |
| Evo-MedAgentModel=gpt-oss-120b2026.05 | 65.3 | |
| No skillsModel=GPT-5.4 (low)2026.05 | 63.5 | |
| Evo-MedAgentModel=Gemini 3.1 Flash Lite2026.05 | 62.9 | |
| Batch MemoryModel=GPT-5.4 (low)2026.05 | 62.9 | |
| Batch MemoryModel=gpt-oss-120b2026.05 | 62.2 | |
| Seq. MemoryModel=GPT-5.4 (low)2026.05 | 62.1 | |
| Seq. MemoryModel=Gemini 3.1 Flash Lite2026.05 | 60 | |
| ExpeLModel=Gemini 3.1 Flash Lite2026.05 | 58.3 | |
| No skillsModel=gpt-oss-120b2026.05 | 57.6 | |
| Batch MemoryModel=Gemini 3.1 Flash Lite2026.05 | 55.4 | |
| Seq. MemoryModel=gpt-oss-120b2026.05 | 54.2 | |
| No skillsModel=Gemini 3.1 Flash Lite2026.05 | 47.6 | |
| GRASPModel=DeepSeek V4 Flash2026.05 | 36.8 | |
| ExpeLModel=DeepSeek V4 Flash2026.05 | 31.5 | |
| Evo-MedAgentModel=DeepSeek V4 Flash2026.05 | 25.5 | |
| No skillsModel=DeepSeek V4 Flash2026.05 | 23.8 | |
| SkillXModel=DeepSeek V4 Flash2026.05 | 19.5 | |
| Seq. MemoryModel=DeepSeek V4 Flash2026.05 | 17.8 | |
| Batch MemoryModel=DeepSeek V4 Flash2026.05 | 17.8 |