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

OmniVideoBench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Video UnderstandingOmniVideoBench
Score40.3
50
Audio-Video UnderstandingOmniVideoBench
Avg Latency29.2
23
Audio-Visual Video UnderstandingOmniVideoBench truncated <= 5min
Score57.6
20
Omnimodal Question AnsweringOmniVideoBench 1.0 (test)
Compare Attr44.44
18
Audio-visual Question AnsweringOmniVideoBench
Accuracy0.356
18
Fine-grained audio-visual video understandingOmniVideoBench
Accuracy58.9
14
Audio-Visual Joint ReasoningOmniVideoBench
Music Score56.2
11
Long-form audio-video understandingOmniVideoBench up to 30min
Latency (10, 30] min26.2
9
Video ReasoningOmniVideoBench
Accuracy (Long)40.52
8
Omni-modal collaborative reasoningOmniVideoBench
Top-1 Accuracy40.5
6
Showing 10 of 10 rows