VideoMME w/o sub.
multimodal
+
+
+
+
About
VideoMME w/o Sub is the subtitle-free variant of the VideoMME benchmark, testing AI models' video understanding capabilities without textual assistance from subtitles or captions. This challenging evaluation focuses purely on visual and audio analysis, assessing models' ability to comprehend video content, answer questions, and perform reasoning tasks through multimodal understanding without linguistic cues.
+
+
+
+
Evaluation Stats
Total Models3
Organizations1
Verified Results0
Self-Reported3
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers
Score Distribution
3 models
Top Score
73.3%
Average Score
69.6%
High Performers (80%+)
0Top Organizations
#1Alibaba Cloud / Qwen Team
3 models
69.6%
+
+
+
+
Leaderboard
3 models ranked by performance on VideoMME w/o sub.
| License | Links | ||||
|---|---|---|---|---|---|
| Jan 26, 2025 | tongyi-qianwen | 73.3% | |||
| Feb 28, 2025 | Apache 2.0 | 70.5% | |||
| Jan 26, 2025 | Apache 2.0 | 65.1% |
