Sokoban on Sokoban (Hard-1)
91.41AccuracyGPT-5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5category=Proprietary Models2025.11 | 91.41 | |
| OpenAI o3category=Proprietary Models2025.11 | 90.23 | |
| OpenAI o4-minicategory=Proprietary Models2025.11 | 69.14 | |
| WMAct2025.11 | 52.68 | |
| PPO - Interactivemode=interactive2025.11 | 41.26 | |
| Claude 4.5 Sonnetcategory=Proprietary Models2025.11 | 34.77 | |
| Gemini 2.5 Procategory=Proprietary Models2025.11 | 17.19 | |
| Qwen3-14Bcategory=Opensource Models2025.11 | 6.64 | |
| Qwen3-8Bcategory=Opensource Models2025.11 | 4.6 | |
| PPO - EntirePlanmode=single-turn output2025.11 | 2.34 | |
| Qwen3-8B-Ownbackbone=Qwen3-8B2025.11 | 0.84 | |
| GPT-4ocategory=Proprietary Models2025.11 | 0 | |
| Qwen2.5-7B-Instructcategory=Opensource Models2025.11 | 0 | |
| Qwen2.5-32B-Instructcategory=Opensource Models2025.11 | 0 |