Long-context modeling
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
9Decoding Speedup
13
Jun 2, 2026
3.85Speedup
12
Jun 2, 2026
96.74Accuracy (4K Context)
10
Feb 26, 2026
50.5LCC
6
May 8, 2026
90.2NIAH Score (MK, 1K)
6
Apr 23, 2026
23.92S. QA Accuracy
5
Apr 15, 2026
9Decoding Speedup
1
Apr 1, 2026