Maze Navigation on Maze (Standard)
0.9961AccuracyGPT-5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5category=Proprietary Models2025.11 | 0.9961 | |
| OpenAI o3category=Proprietary Models2025.11 | 0.957 | |
| Claude 4.5 Sonnetcategory=Proprietary Models2025.11 | 0.9414 | |
| OpenAI o4-minicategory=Proprietary Models2025.11 | 0.9375 | |
| WMAct2025.11 | 0.8814 | |
| PPO - Interactivemode=interactive2025.11 | 0.8374 | |
| Gemini 2.5 Procategory=Proprietary Models2025.11 | 0.8359 | |
| PPO - EntirePlanmode=single-turn output2025.11 | 0.7504 | |
| Qwen3-14Bcategory=Opensource Models2025.11 | 0.7227 | |
| Qwen3-8Bcategory=Opensource Models2025.11 | 0.6367 | |
| GPT-4ocategory=Proprietary Models2025.11 | 0.0469 | |
| Qwen3-8B-Ownbackbone=Qwen3-8B2025.11 | 0.0195 | |
| Qwen2.5-32B-Instructcategory=Opensource Models2025.11 | 0.0068 | |
| Qwen2.5-7B-Instructcategory=Opensource Models2025.11 | 0 |