List your company

General Embodied AI Performance on Aggregate Sokoban, FrozenLake, Navigation, PrimitiveSkill

93Overall Success RateAtlasVA
5.6428.325173.68May 18, 2026

Evaluation results

MethodLinks
AtlasVAModel Category=Ours, Parameter Count=3B2026.0593
VAGENModel Category=Open-Source2026.0578
o3Model Category=Proprietary2026.0571
GPT-5Model Category=Proprietary2026.0569
Claude Sonnet 4.5Model Category=Proprietary2026.0562
o4-miniModel Category=Proprietary2026.0560
GPT-4oModel Category=Proprietary2026.0560
Gemini 2.5 flashModel Category=Proprietary2026.0558
Qwen2.5-VL-72BModel Category=Open-Source2026.0555
Gemini 2.5 ProModel Category=Proprietary2026.0551
Claude Sonnet 3.7Model Category=Proprietary2026.0551
Gemini 2.0Model Category=Proprietary2026.0539
Qwen2.5-VL-7BModel Category=Open-Source2026.0519
VLM-R1-3BModel Category=Open-Source2026.0510
Qwen2.5-VL-3BModel Category=Open-Source2026.059