VideoOdyssey-AV: leaderboard

VideoOdyssey (2026) tests ultra-long video understanding with human-written multiple-choice questions on videos averaging 109 minutes, each annotated with how long a viewer must watch continuously to answer. This board is accuracy in percent over all questions of the audio-visual subset, which need both picture and sound; higher is better.

Metric: Multiple-choice accuracy (%) over all questions 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 June 2027. 12 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Pro (Preview)46.1#99
2Gemini 3 Flash44.3#175
3Gemini 2.5 Pro43.9#276
4Qwen 3.5 Omni Plus43#254
5Ola-7B31.5#1557
6Qwen 3 Omni 30B A3B28.7#847
7Unified-IO-2 XL26.4
8Unified-IO-2 XXL25.9
9Unified-IO-2 L25.1
10VideoLLaMA2-7B24.1#2385
11VITA-1.5-7B22.6#1853
12OneLLM21.2#2372

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=videoodyssey-av · How It Works · Data refreshed daily, snapshot 2026-11-21.