VideoOdyssey-AV - Captioning: leaderboard
Metric: Multiple-choice accuracy (%) on the 47 captioning questions (generate a descriptive text that details the observable visual actions and audible acoustic events, including speech, environmental sounds, and scene dynamics, occurring in the video) of VideoOdyssey-AV (human-written questions on 100 videos of 60 to 255 minutes that need both visual and audio cues; 1,062 in all), omni-modal models given video and audio, 64 frames (VITA-1.5-7B 16); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 74.5 |
| 2 | Qwen 3.5 Omni Plus | 72.3 |
| 3 | Gemini 3 Flash | 66 |
| 4 | Gemini 3.1 Pro (Preview) | 61.7 |
Interactive version: theaggregate.ai/benchmark?slug=videoodyssey-av-captioning · How It Works · Data refreshed daily, snapshot 2026-10-07.