LongVideoBench
multimodal
+
+
+
+
About
LongVideoBench is a video-language benchmark featuring hour-long interleaved inputs that evaluate multimodal models' ability to understand extended video content. With 3,763 web-collected videos across diverse themes, this benchmark tests models' capacity for long-context video comprehension, temporal reasoning, and sustained attention when processing extended multimodal sequences with subtitles and visual information.
+
+
+
+
Evaluation Stats
Total Models1
Organizations1
Verified Results0
Self-Reported1
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers
Score Distribution
1 models
Top Score
54.7%
Average Score
54.7%
High Performers (80%+)
0Top Organizations
#1Alibaba Cloud / Qwen Team
1 model
54.7%
+
+
+
+
Leaderboard
1 models ranked by performance on LongVideoBench
| License | Links | ||||
|---|---|---|---|---|---|
| Jan 26, 2025 | Apache 2.0 | 54.7% |
