List your company

Embodied Navigation on EB-Navigation

85Average ScoreGPT-5
22.558438.769254.9871.1908May 26, 2026Jun 1, 2026Jun 7, 2026Jun 13, 2026Jun 19, 2026Jun 25, 2026Jul 1, 2026

Evaluation results

MethodLinks
GPT-5Tool use=Yes2026.05858587858385
Claude-4.6Tool use=Yes2026.05858585858585
Qwen3-32BTool use=Yes2026.05848382858287
GPT-4oTool use=Yes2026.05838085858083
Claude-3.7Tool use=Yes2026.05838582838582
Qwen2.5-32BTool use=Yes2026.05817583838283
Qwen3.5-35BTool use=Yes2026.05818080808382
Qwen3-8BTool use=Yes2026.05808585777577
Claude-4.6Tool use=No2026.05626863775548
MuSixbackbone=Qwen3-VL-4B-Instruct, world model groups (G)=3, world models (K)=15, top-k models=4, alpha_max=0.3, gamma=1.02026.0757.9263.263.257.662.443.2
GPT-5Tool use=No2026.05576262685042
Claude-3.7Tool use=No2026.05556762635035
SayCanPay2026.0747.254.447.257.64828.8
GPT-4oTool use=No2026.05465850473343
Qwen3-8BTool use=No2026.05426248453222
Qwen3-32BTool use=No2026.05405247553510
Qwen3.5-35BTool use=No2026.0539504248477
Qwen2.5-32BTool use=No2026.05354842333023
FLAREparadigm=retrieval-augmented generation2026.0734.2445.635.2443214.4
Conventional MoErouting=standard top-k routing2026.0729.283628.839.232.89.6
LLM-Plannerparadigm=few-shot prompting without explicit world modeling2026.0724.9636.830.427.228.81.6