AudioMC: leaderboard
AudioMultiChallenge benchmarks E2E spoken dialogue systems on multi-turn interaction, voice editing, and instruction retention.
Metric: Score (imported). Source: labs.scale.com. Source: benchmarklist.com. Saturation forecast: Around September 2027. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) (High) | 66.8 | #58 (Gemini 3.1 Pro (Preview)) |
| 2 | Inkling | 56.6 | #126 |
| 3 | Inkling Small | 54.9 | #138 |
| 4 | Gemini 2.5 Pro (Thinking) | 46.9 | #152 (Gemini 2.5 Pro) |
| 5 | Gemini 2.5 Flash (Thinking) | 40.04 | #245 (Gemini 2.5 Flash) |
| 6 | GPT Realtime 2 | 37.61 | #133 |
| 7 | Voxtral-Small-24B-2507 | 26.33 | #581 |
| 8 | Gemini 2.5 Flash | 26.11 | #245 |
| 9 | Qwen3 Omni 30B A3B Instruct | 24.34 | #383 |
| 10 | Phi-4 Multimodal Instruct | 15.49 | #906 |
| 11 | Gemma 3n E4B (IT) | 15.49 | #749 |
| 12 | Qwen2.5-Omni-7B | 11.95 | #624 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=audiomc · How It Works · Data refreshed daily, snapshot 2026-10-20.