Complex Reasoning QA on SpeechMMLU (5-shot)
69.1S2S AccuracyMiMo-Audio
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MiMo-AudioModel Scale=7B-Base, Few-shot setting=5-shot2025.12 | 69.1 | 69.5 | 71.5 | 72.5 | |
| Step-Audio2-miniModel Scale=7B-Base, Few-shot setting=5-shot2025.12 | 51.8 | 67.8 | 63.4 | 74.1 | |
| Baichuan-AudioModel Scale=7B-Base, Few-shot setting=5-shot2025.12 | 31.9 | 29.9 | 16.7 | 71.1 | |
| Kimi-AudioModel Scale=7B-Base, Few-shot setting=5-shot2025.12 | 11.8 | 67.9 | 0 | 70.7 |