Agentic Planning and Ideation on MUTATE (test)
41Success Count (Path Discovery)ReDNA
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ReDNAModel=Claude Sonnet 4.62026.05 | 41 | 32.8 | 2.616 | 3.205 | 3.533 | |
| ReDNAModel=GPT-5.42026.05 | 38 | 40.2 | 2.771 | 3.262 | 3.725 | |
| EscapeAgentModel=Claude Sonnet 4.62026.05 | 37 | 43.9 | 2.526 | 2.743 | 3.341 | |
| ReDNAModel=Qwen3 235B A22B2026.05 | 36 | 36.5 | 2.707 | 3.407 | 3.353 | |
| EscapeAgentModel=GPT-5.42026.05 | 35 | 44.9 | 2.542 | 2.99 | 3.516 | |
| EscapeAgentModel=Qwen3 235B A22B2026.05 | 34 | 48.3 | 2.645 | 3.148 | 3.47 | |
| BaseModel=Claude Sonnet 4.62026.05 | 30 | 26.9 | 2.764 | 3.079 | 3.35 | |
| Self-RefineModel=GPT-5.42026.05 | 29 | 33 | 2.738 | 2.703 | 3.514 | |
| ReDNAModel=Llama 4 Maverick2026.05 | 29 | 45.3 | 2.409 | 3.021 | 3.031 | |
| BaseModel=GPT-5.42026.05 | 27 | 26.3 | 2.647 | 2.827 | 3.588 | |
| Self-RefineModel=Qwen3 235B A22B2026.05 | 27 | 35.7 | 2.724 | 3.181 | 3.449 | |
| Self-RefineModel=Claude Sonnet 4.62026.05 | 25 | 21.9 | 2.733 | 3.088 | 3.457 | |
| EscapeAgentModel=Llama 4 Maverick2026.05 | 23 | 52.7 | 2.214 | 2.571 | 3.11 | |
| BaseModel=Qwen3 235B A22B2026.05 | 22 | 33.8 | 2.755 | 3.115 | 3.388 | |
| BaseModel=Llama 4 Maverick2026.05 | 13 | 36.2 | 2.256 | 2.667 | 3.128 | |
| Self-RefineModel=Llama 4 Maverick2026.05 | 13 | 31.2 | 2.324 | 2.681 | 3.181 |