ResearchTasksReal-World AgentReal-World AgentFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedClaw-EvalGPT-5.480.6Average Score22Jun 2, 2026PinchBenchClaude Opus 4.682.3Average Score15Jun 2, 2026