Long-context retrieval and aggregation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
89.56Average Accuracy
76
May 27, 2026
93.07Average Accuracy
76
May 27, 2026
94.24Average Accuracy
76
May 27, 2026
94.73Average Accuracy
76
May 27, 2026
95.6NIAH-MV
3
May 12, 2026
95.4NIAH-MV
3
May 12, 2026
98.9NIAH-MV
3
May 12, 2026
37.2Performance Score (32k Context)
2
Jun 9, 2026
48.7Performance (32k Context)
2
Jun 9, 2026