List your company

Reward Modeling on OSWorld Verified Class-Balanced Human Evaluation 1.0 (test)

94PrecisionVAGEN
85.36887.60989.8592.091Jan 31, 2026

Evaluation results

MethodLinks
VAGENVerified Agent Backbone=Claude-Sonnet-4.52026.019495.294.692.9
AndroidGenVerified Agent Backbone=Claude-Sonnet-4.52026.0192.67784.180.2
WebRLVerified Agent Backbone=Claude-Sonnet-4.52026.0191.677.283.879.8
DistRLVerified Agent Backbone=Claude-Sonnet-4.52026.0187.774.280.475.6
FullTrajEvalVerified Agent Backbone=Claude-Sonnet-4.52026.0187.789.588.684.7
ZeroGUIVerified Agent Backbone=Claude-Sonnet-4.52026.0187.188.287.783.7
DigiRLVerified Agent Backbone=Claude-Sonnet-4.52026.0185.772.878.773.8