List your company

Spatial Planning on Sokoban

79Success RateAtlasVA
9.8427.79545.7563.705May 14, 2026May 15, 2026May 16, 2026May 17, 2026May 18, 2026

Evaluation results

MethodLinks
AtlasVAModel Category=Ours, Parameter Count=3B2026.0579
GPT-5Model Category=Proprietary2026.0570
VAGENModel Category=Open-Source2026.0561
o3Model Category=Proprietary2026.0560
Gemini 2.5 ProModel Category=Proprietary2026.0558
Gemini 2.5 flashModel Category=Proprietary2026.0552
o4-miniModel Category=Proprietary2026.0544
GPT-4oModel Category=Proprietary2026.0543
ActFocusBackbone=3B, Evaluation Step=2002026.0537.5
ActFocusBackbone=0.5B, Evaluation Step=2002026.0534
Claude Sonnet 4.5Model Category=Proprietary2026.0531
Gemini 2.0Model Category=Proprietary2026.0528
UniformBackbone=3B, Evaluation Step=2002026.0526.2
Claude Sonnet 3.7Model Category=Proprietary2026.0525
Qwen2.5-VL-72BModel Category=Open-Source2026.0518
Qwen2.5-VL-3BModel Category=Open-Source2026.0514
Qwen2.5-VL-7BModel Category=Open-Source2026.0513
VLM-R1-3BModel Category=Open-Source2026.0513
UniformBackbone=0.5B, Evaluation Step=2002026.0512.5