Omni-Persona - Audio-to-Audio: leaderboard
Metric: Calibrated accuracy (%): item-weighted accuracy over answerable items (answer judged correct by a GPT-5.4-mini judge) and absent-persona items (correct abstention by keyword matching), on the audio-to-audio persona matching group (query voice clip, personal audio context), zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 4 E2B (IT) | 31.2 |
| 2 | Gemini 3.6 Flash (Medium) | 23.9 |
| 3 | Phi-4 Multimodal Instruct | 20.6 |
| 4 | Qwen2.5-Omni-7B | 17 |
| 5 | gemma-4-E4B-it | 16.6 |
| 6 | Gemini 3.6 Flash (High) | 16.2 |
| 7 | Qwen3 Omni 30B A3B Instruct | 15.8 |
| 8 | Gemini 3.5 Flash Lite | 10.1 |
Interactive version: theaggregate.ai/benchmark?slug=omni-persona-audio-to-audio · How It Works · Data refreshed daily, snapshot 2026-10-07.