AudioProcessBench - Majority Voting Selection: leaderboard

Metric: Final-answer accuracy (%) when the critic's chain scores are summed per candidate answer; 3,872 reasoning traces (23,497 steps, 41% erroneous) from six audio and omni models on MMAR, MMSU and MMAU-Pro questions, steps labelled by two LLM annotators with human review; few-shot critic prompt with three annotated example chains, temperature 0.7; random 40.6; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 11 models tracked.

Top models

#ModelScore
1Gemini 3 Flash61.5
2Gemma 3n E4B54
3Phi-4 Multimodal Instruct53.7
4Gemma 4 E2B53.2
5Qwen2.5-Omni-7B53.1
6Gemma 3n E2B52.5
7Gemma 4 E4B52.2

Interactive version: theaggregate.ai/benchmark?slug=audioprocessbench-majority-voting-selection · How It Works · Data refreshed daily, snapshot 2026-09-29.