AVI-Bench - Audio-Visual Captioning: leaderboard

Metric: Task score (%) on audio-visual captioning (280 videos): FENSE, the best sentence-BERT similarity to the reference captions with a fluency-error penalty, on AVI-Bench (5,864 audio-visual items in 14 tasks over four cognitive stages), zero-shot with the same prompts for every omni-modal model; task scores normalized to percentages; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 26 models tracked.

Top models

#ModelScore
1Gemini 2.5 Flash (Preview 04-17)57.97
2Qwen2.5-Omni-7B57.22
3Gemini 2.5 Pro (Preview 05-06)56.46
4Gemini 2.0 Flash56.31
5Gemini 1.5 Pro55.03
6Gemini 1.5 Flash54
7Phi-4 Multimodal Instruct51.04

Interactive version: theaggregate.ai/benchmark?slug=avi-bench-audio-visual-captioning · How It Works · Data refreshed daily, snapshot 2026-09-29.