Clinical Task Execution on MedAgentBench (test)
88.8AccuracyGRASP
Evaluation Results
| Method | Links | |
|---|---|---|
| GRASPModel=gpt-oss-120b2026.05 | 88.8 | |
| GRASPModel=GPT-5.4 (low)2026.05 | 85.4 | |
| GRASPModel=GPT-4.12026.05 | 84.9 | |
| GRASPModel=Gemini 3.1 Flash Lite2026.05 | 71.4 | |
| GRASPModel=DeepSeek V4 Flash2026.05 | 70 | |
| Evo-MedAgentModel=gpt-oss-120b2026.05 | 67.8 | |
| SkillXModel=DeepSeek V4 Flash2026.05 | 55.9 | |
| ExpeLModel=DeepSeek V4 Flash2026.05 | 55.6 | |
| Evo-MedAgentModel=Gemini 3.1 Flash Lite2026.05 | 55.2 | |
| No skillsModel=Gemini 3.1 Flash Lite2026.05 | 54.2 | |
| Seq. MemoryModel=Gemini 3.1 Flash Lite2026.05 | 54.2 | |
| SkillXModel=Gemini 3.1 Flash Lite2026.05 | 54.2 | |
| ExpeLModel=Gemini 3.1 Flash Lite2026.05 | 53.6 | |
| SkillXModel=gpt-oss-120b2026.05 | 53.1 | |
| Evo-MedAgentModel=DeepSeek V4 Flash2026.05 | 51.9 | |
| ExpeLModel=gpt-oss-120b2026.05 | 49.1 | |
| No skillsModel=DeepSeek V4 Flash2026.05 | 47.7 | |
| Batch MemoryModel=GPT-4.12026.05 | 47.4 | |
| Seq. MemoryModel=GPT-5.4 (low)2026.05 | 47.4 | |
| Batch MemoryModel=GPT-5.4 (low)2026.05 | 46.4 | |
| No skillsModel=GPT-4.12026.05 | 45.5 | |
| Seq. MemoryModel=GPT-4.12026.05 | 45.3 | |
| ExpeLModel=GPT-4.12026.05 | 45.3 | |
| Evo-MedAgentModel=GPT-4.12026.05 | 45.3 | |
| No skillsModel=GPT-5.4 (low)2026.05 | 45.1 | |
| SkillXModel=GPT-5.4 (low)2026.05 | 44.8 | |
| ExpeLModel=GPT-5.4 (low)2026.05 | 43.8 | |
| Evo-MedAgentModel=GPT-5.4 (low)2026.05 | 43.8 | |
| Seq. MemoryModel=gpt-oss-120b2026.05 | 41.2 | |
| No skillsModel=gpt-oss-120b2026.05 | 40.6 | |
| Batch MemoryModel=Gemini 3.1 Flash Lite2026.05 | 39.6 | |
| SkillXModel=GPT-4.12026.05 | 38.5 | |
| Batch MemoryModel=DeepSeek V4 Flash2026.05 | 37.8 | |
| Seq. MemoryModel=DeepSeek V4 Flash2026.05 | 35 | |
| Batch MemoryModel=gpt-oss-120b2026.05 | 34.4 |