AVI-Bench - Audio Sensation QA: leaderboard

Metric: Task score (%) on primitive audio sensation (502 multiple-choice items on unfamiliar, low-semantic synthetic sounds): accuracy under double confirmation, 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 Pro (Preview 05-06)29.48
2Gemini 2.5 Flash (Preview 04-17)23.11
3Gemini 1.5 Flash21.91
4Gemini 2.0 Flash21.51
5Qwen2.5-Omni-7B21.12
6Gemini 1.5 Pro17.93
7Phi-4 Multimodal Instruct1.59

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