List your company

Embodied Household Navigation on ALFWorld

97Success RateGemini-3-Flash
1939.2559.579.75May 13, 2026

Evaluation results

MethodLinks
Gemini-3-FlashProtocol=Prompting2026.0597
ICRLBackbone=Qwen3-4B2026.0592.5
Critique-GRPOBackbone=Qwen3-4B2026.0591.5
ScalingInterBackbone=Qwen3-8B2026.0591
Critique-GRPOBackbone=Qwen3-8B2026.0590.5
ICRLBackbone=Qwen3-8B2026.0590.5
ScalingInterBackbone=Qwen3-4B2026.0590
MATPOBackbone=Qwen3-8B2026.0589
MATPOBackbone=Qwen3-4B2026.0587.5
GSPOBackbone=Qwen3-8B2026.0583
GRPOBackbone=Qwen3-8B2026.0582.5
GSPOBackbone=Qwen3-4B2026.0582
GRPOBackbone=Qwen3-4B2026.0581
Gemini-2.5-flashProtocol=Prompting2026.0578.5
Qwen3-30B-A3BProtocol=Prompting2026.0545.5
Qwen3-8BProtocol=Prompting2026.0538
Qwen3-4BProtocol=Prompting2026.0522