Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Long Video Understanding on LongVideoBench (test)

66.7Accuracy

GPT-4o

51.51655.45859.463.342Jan 4, 2026Jan 26, 2026Feb 18, 2026Mar 13, 2026Apr 5, 2026Apr 28, 2026May 21, 2026
Updated 2mo ago

Evaluation Results

MethodLinks
2026.01
66.7
2026.05
60.1
2026.03
58.9
2026.01
58.8
2026.01
58.7
2026.05
58.5
2026.05
58.2
2026.05
57.7
2026.03
57.3
2026.03
57.1
2026.03
57.1
2026.01
56.8
2026.05
56.7
2026.05
56.5
2026.03
56.3
2026.05
56.3
2026.03
54.8
2026.03
54.2
2026.05
52.1