ResearchTasksLong-horizon memory Question AnsweringLong-horizon memory Question AnsweringFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedLongBenchGPT-4o (ceiling)73.7Accuracy9May 6, 2026LongMemEvalGPT-4o (ceiling)68.4Accuracy9May 6, 2026ATM-Bench subset431Graph-NoReader69.61EM3May 29, 2026