ResearchTasksHuman Preference AgreementHuman Preference AgreementFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedMM-RewardBench2 T2IGemini 3.1 Pro + ARR78.9Accuracy13May 12, 2026MM-RewardBench2 EditGemini 3.1 Pro + ARR79.2Accuracy7May 12, 2026