AOR-Bench - MB-Score: leaderboard
Metric: MB-Score (%; harmonic mean of the true-refusal rate on 500 harmful speech-only queries and 100 minus the over-refusal rate, pooled over six scenario categories). Source: arxiv.org. Saturation forecast: Around 2028. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen2-Audio-7B-Instruct | 82.8 |
| 2 | Gemini 3 Flash (Preview) | 55.9 |
| 3 | MiMo-V2-Omni | 55.68 |
| 4 | Gemini 2.0 Flash Lite | 54.16 |
| 5 | GPT Audio | 48.17 |
| 6 | Gemini 2.5 Flash Lite | 47.97 |
| 7 | GPT Audio Mini | 39.22 |
Interactive version: theaggregate.ai/benchmark?slug=aor-bench-mb-score · How It Works · Data refreshed daily, snapshot 2026-09-26.