VideoOdyssey-V - Captioning: 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 on the visual subset questions about describing a scene; higher is better.

Metric: Multiple-choice accuracy (%) on the 49 captioning questions (generate a text description that details the specific visual actions, objects, and scene dynamics observable in the video) of VideoOdyssey-V (human-written questions on 100 videos of 60 to 255 minutes that need visual cues and long continuous viewing; 1,618 in all), 128 frames for proprietary models and each open model's maximum frame setting; higher is better. Source: arxiv.org. Saturation forecast: Around October 2027. 18 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Pro (Preview)59.2#99
2Seed 2.0 Pro57.1#222
3Qwen 3.5 27B51#375
4GPT-5.249#185
5Gemini 2.5 Pro44.9#276
6Kimi K2.544.9#266
7Qwen 3 VL 235B A22B44.9#485
8InternVL3.5-38B42.9#1021
9Qwen 3 VL 32B38.8#622
10Video-KTR-7B38.8
11Video-LLaVA-7B36.7#3879
12Video-R136.7#1379
13LLaVA-NeXT-Video-DPO (34B)34.7#2489
14Phi-4 Multimodal Instruct32.7#4154
15LLaVA-OneVision-1.5-8B32.7#1864

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

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