Closed-world plan-synthesis on Blocksworld Mystery variant
100Success RateFast Downward
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Fast Downwardupper bound=true2026.07 | 100 | — | |
| EvoPlanbackbone=gpt-oss-120b2026.07 | 76.7 | 363,100 | |
| EvoPlanbackbone=Qwen3-32B2026.07 | 74 | 1,690,032 | |
| NL-PDDL2026.07 | 70 | — | |
| Gemini-2.0 Flash2026.07 | 1 | 834,060 | |
| GPT-4o2026.07 | 0 | 834,890 | |
| LLaMA-3.12026.07 | 0 | 841,791 |