List your company

World Modelling on AlfWorld

61.5Exact Matchgemini-3-flash-preview-thinking
19.58830.46941.3552.231Jun 16, 2026

Evaluation results

MethodLinks
gemini-3-flash-preview-thinkingTask Type=clean2026.0661.588.279.990.5
Looped World Model (Ours)Task Type=look, Number of parameters=about 1B2026.0660.578.973.182
claude-opus-4-6-maxTask Type=look2026.0660.573.268.978.6
Looped World Model (Ours)Task Type=clean, Number of parameters=about 1B2026.0660.481.77581.3
claude-opus-4-6-maxTask Type=clean2026.0657.373.868.977.4
gemini-3-flash-preview-thinkingTask Type=look2026.0655.886.674.497.7
Looped World Model (Ours)Task Type=heat, Number of parameters=about 1B2026.065581.876.681.2
gemini-3-flash-preview-thinkingTask Type=heat2026.065581.973.390.6
gemini-3-flash-preview-thinkingTask Type=cool2026.0654.588.17788.3
claude-opus-4-6-maxTask Type=Overall2026.065372.666.877
claude-opus-4-6-maxTask Type=heat2026.0652.567.664.474.1
Looped World Model (Ours)Task Type=Overall, Number of parameters=about 1B2026.0651.680.471.681.1
claude-opus-4-6-maxTask Type=pick2026.065172.865.778.1
Looped World Model (Ours)Task Type=cool, Number of parameters=about 1B2026.065081.772.678.5
claude-opus-4-6-maxTask Type=cool2026.06507368.272.8
gemini-3-flash-preview-thinkingTask Type=Overall2026.065083.57190.2
Looped World Model (Ours)Task Type=pick, Number of parameters=about 1B2026.0646.779.769.181.5
gemini-3-flash-preview-thinkingTask Type=pick2026.0642.780.265.189.3
qwen-3.5-flashTask Type=clean2026.0636.571.955.690.1
qwen-3.5-flashTask Type=look2026.0627.966.246.392.8
qwen-3.5-flashTask Type=heat2026.0627.570.149.591.2
qwen-3.5-flashTask Type=cool2026.0627.37252.685.1
qwen-3.5-flashTask Type=Overall2026.062667.347.788.4
qwen-3.5-flashTask Type=pick2026.0621.264.143.587.5