List your company

World Modelling (QA/Text Metrics) on ScienceWorld

100EM (Exact Match)Looped World Model
-4235077Jun 16, 2026

Evaluation results

MethodLinks
Looped World ModelTask Type=Lifespan, Number of Parameters=1B, Number of consecutive actions=52026.06100100100100——
Looped World ModelTask Type=Conductivity, Number of Parameters=1B, Number of consecutive actions=52026.06878987.887.9——
Looped World ModelTask Type=Thermometer, Number of Parameters=1B, Number of consecutive actions=52026.0683.385.383.383.3——
claude-opus-4-6-maxTask Type=Thermometer, Number of consecutive actions=52026.0683.398.193.597.2——
Looped World ModelTask Type=StateChange, Number of Parameters=1B, Number of consecutive actions=52026.068083.18080——
Looped World ModelTask Type=Genetics, Number of Parameters=1B, Number of consecutive actions=52026.0678.380.278.979.8——
Looped World ModelTask Type=Find, Number of Parameters=1B, Number of consecutive actions=52026.0676.990.482.785.8——
Looped World ModelTask Type=Grow, Number of Parameters=1B, Number of consecutive actions=52026.0673.88075.579.8——
Looped World ModelTask Type=Melt, Number of Parameters=1B, Number of consecutive actions=52026.067391.985.791.6——
claude-opus-4-6-maxTask Type=Grow, Number of consecutive actions=52026.0670.881.676.180.7——
claude-opus-4-6-maxTask Type=Find, Number of consecutive actions=52026.0669.283.878.984.6——
Looped World ModelTask Type=Overall, Number of Parameters=1B, Number of consecutive actions=52026.0668.4——83.985.380.7
Looped World ModelTask Type=Boil, Number of Parameters=1B, Number of consecutive actions=52026.0666.77975.377.5——
Looped World ModelTask Type=Incline, Number of Parameters=1B, Number of consecutive actions=52026.0659.395.390.493.4——
claude-opus-4-6-maxTask Type=Genetics, Number of consecutive actions=52026.0659.271.865.771.6——
Looped World ModelTask Type=Power, Number of Parameters=1B, Number of consecutive actions=52026.0657.163.960.861.5——
claude-opus-4-6-maxTask Type=Conductivity, Number of consecutive actions=52026.0647.867.253.172.7——
claude-opus-4-6-maxTask Type=Overall, Number of consecutive actions=52026.0647.2——72.372.864.4
Looped World ModelTask Type=Chemistry, Number of Parameters=1B, Number of consecutive actions=52026.0644.464.454.257.9——
claude-opus-4-6-maxTask Type=Chemistry, Number of consecutive actions=52026.0644.459.846.359.2——
claude-opus-4-6-maxTask Type=Power, Number of consecutive actions=52026.0642.947.345.345.8——
claude-opus-4-6-maxTask Type=StateChange, Number of consecutive actions=52026.064065.544.673.3——
claude-opus-4-6-maxTask Type=Melt, Number of consecutive actions=52026.0636.552.446.353.4——
claude-opus-4-6-maxTask Type=Incline, Number of consecutive actions=52026.063486.576.283.8——
Looped World ModelTask Type=Freeze, Number of Parameters=1B, Number of consecutive actions=52026.062559.731.254.8——
claude-opus-4-6-maxTask Type=Boil, Number of consecutive actions=52026.0622.233.330.232.3——
claude-opus-4-6-maxTask Type=Freeze, Number of consecutive actions=52026.0612.533.621.637.3——
Looped World ModelTask Type=LifeStages, Number of Parameters=1B, Number of consecutive actions=52026.06018.36.110.2——
claude-opus-4-6-maxTask Type=LifeStages, Number of consecutive actions=52026.06010.666——
claude-opus-4-6-maxTask Type=Lifespan, Number of consecutive actions=52026.06061.4058.3——