ResearchDatasetsAgentRewardBenchAgentRewardBenchFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsReward ModelingAgentRewardBench83.8Precision12Web agent meta-evaluationAGENTREWARDBENCH89.21Overall Precision8Reward VerificationAgentRewardBench VisualWebArena100Precision7