2D discrete grid-world planning on FrozenLake
83Success RateAtlasVA
Evaluation Results
| Method | Links | |
|---|---|---|
| AtlasVAModel Category=Ours, Parameter Count=3B2026.05 | 83 | |
| o4-miniModel Category=Proprietary2026.05 | 82 | |
| Claude Sonnet 4.5Model Category=Proprietary2026.05 | 80 | |
| o3Model Category=Proprietary2026.05 | 78 | |
| Gemini 2.5 ProModel Category=Proprietary2026.05 | 78 | |
| GPT-5Model Category=Proprietary2026.05 | 77 | |
| VAGENModel Category=Open-Source2026.05 | 71 | |
| Claude Sonnet 3.7Model Category=Proprietary2026.05 | 69 | |
| Gemini 2.0Model Category=Proprietary2026.05 | 61 | |
| Gemini 2.5 flashModel Category=Proprietary2026.05 | 57 | |
| GPT-4oModel Category=Proprietary2026.05 | 54 | |
| Qwen2.5-VL-72BModel Category=Open-Source2026.05 | 44 | |
| Qwen2.5-VL-7BModel Category=Open-Source2026.05 | 14 | |
| Qwen2.5-VL-3BModel Category=Open-Source2026.05 | 14 | |
| VLM-R1-3BModel Category=Open-Source2026.05 | 13 |