ResearchTasksCertified Task-solvingCertified Task-solvingFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedPHACT Safety Suite (Impossible goals)Llama 3.31False Certifications (Bare)2Jun 30, 2026PHACT Safety Suite Feasible goalsGemini 2.552Bare Feasible Valid Score2Jun 30, 2026