Clinical Task Execution on MedAgentBench (val)
86.2AccuracyGRASP
Evaluation Results
| Method | Links | |
|---|---|---|
| GRASPModel=GPT-4.12026.05 | 86.2 | |
| GRASPModel=gpt-oss-120b2026.05 | 86 | |
| GRASPModel=GPT-5.4 (low)2026.05 | 84.6 | |
| GRASPModel=DeepSeek V4 Flash2026.05 | 68.8 | |
| GRASPModel=Gemini 3.1 Flash Lite2026.05 | 66.7 | |
| Evo-MedAgentModel=gpt-oss-120b2026.05 | 62.5 | |
| SkillXModel=DeepSeek V4 Flash2026.05 | 55 | |
| SkillXModel=Gemini 3.1 Flash Lite2026.05 | 54.2 | |
| Evo-MedAgentModel=DeepSeek V4 Flash2026.05 | 52.5 | |
| SkillXModel=gpt-oss-120b2026.05 | 51.5 | |
| ExpeLModel=Gemini 3.1 Flash Lite2026.05 | 51.2 | |
| ExpeLModel=DeepSeek V4 Flash2026.05 | 50.7 | |
| Evo-MedAgentModel=Gemini 3.1 Flash Lite2026.05 | 49.6 | |
| No skillsModel=Gemini 3.1 Flash Lite2026.05 | 49 | |
| Seq. MemoryModel=Gemini 3.1 Flash Lite2026.05 | 48.8 | |
| Batch MemoryModel=GPT-5.4 (low)2026.05 | 47.9 | |
| ExpeLModel=gpt-oss-120b2026.05 | 47.5 | |
| Seq. MemoryModel=GPT-5.4 (low)2026.05 | 46.7 | |
| Batch MemoryModel=GPT-4.12026.05 | 46.2 | |
| SkillXModel=GPT-5.4 (low)2026.05 | 45.4 | |
| ExpeLModel=GPT-5.4 (low)2026.05 | 45 | |
| Seq. MemoryModel=GPT-4.12026.05 | 42.5 | |
| ExpeLModel=GPT-4.12026.05 | 42.5 | |
| Evo-MedAgentModel=GPT-4.12026.05 | 42.5 | |
| Evo-MedAgentModel=GPT-5.4 (low)2026.05 | 42.5 | |
| No skillsModel=DeepSeek V4 Flash2026.05 | 41.8 | |
| No skillsModel=GPT-5.4 (low)2026.05 | 40.6 | |
| No skillsModel=GPT-4.12026.05 | 40.1 | |
| Seq. MemoryModel=gpt-oss-120b2026.05 | 38.5 | |
| Batch MemoryModel=gpt-oss-120b2026.05 | 36.8 | |
| Batch MemoryModel=DeepSeek V4 Flash2026.05 | 36.5 | |
| No skillsModel=gpt-oss-120b2026.05 | 36.4 | |
| SkillXModel=GPT-4.12026.05 | 33.8 | |
| Seq. MemoryModel=DeepSeek V4 Flash2026.05 | 33 | |
| Batch MemoryModel=Gemini 3.1 Flash Lite2026.05 | 29.2 |