Long Video Understanding
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
57.7Accuracy
10
May 15, 2026
69.84Accuracy (15s)
10
Mar 6, 2026
43.6Accuracy
9
Apr 21, 2026
65.6Accuracy
9
Feb 26, 2026
66.7Accuracy
8
Jun 29, 2026
68,478.1Average Context Length (tokens)
8
May 27, 2026
55.4MDR
8
May 19, 2026
47.8MDR
8
May 19, 2026
45.1Accuracy
8
Feb 26, 2026
69.8MLVU m-avg Score
7
Jun 29, 2026
62Accuracy
7
Apr 21, 2026
38.6Accuracy (long-acc)
7
Apr 14, 2026
0.021Inference Time (s)
7
Mar 5, 2026
70Accuracy
7
Feb 26, 2026
0.827Retrieval Score
7
Feb 26, 2026
52.7Accuracy
6
Jul 2, 2026
65.5Accuracy
6
Jul 2, 2026
35.9Overall Score
6
May 15, 2026
80.6Accuracy
5
Jun 16, 2026
60.9Accuracy
5
Jun 11, 2026
79.8Accuracy
5
Feb 26, 2026
53.3Overall Accuracy
4
Jul 7, 2026
53.5Base Score
3
Mar 4, 2026
60.4MLVU-Avg
3
Feb 26, 2026
-0.129Pearson's r
1
Feb 26, 2026