Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

WorldArena

Benchmarks

Task NameDataset NameSOTA ResultTrend
World ModelingWorldArena (test)
Image Quality67.36
15
Dual-arm manipulationWorldArena
EWM Score62.84
14
Video GenerationWorldArena
Interaction Quality68.2
14
Action-Conditioned Video GenerationWorldArena
EWMScore70.67
10
Embodied World ModelingWorldArena Robotwin
Interaction Quality Score0.682
9
Human evaluation of robot rollout generationWorldArena rollouts
Task Success1.81
8
Action PlanningWorldArena
Task 1 Closed-Loop Success Rate22
7
World ModelingWorldArena
EWMScore59.7
7
Video PerceptionWorldArena
Img Score0.449
5
Embodied Robotics NavigationWorldArena
2D nDTW9.006
4
Showing 10 of 10 rows