Multimodal Speech Understanding on MMSU All (test)
75.2AccuracyStep-Audio-R1
Evaluation Results
| Method | Links | |
|---|---|---|
| Step-Audio-R12026.05 | 75.2 | |
| Gemini 2.52026.05 | 71.88 | |
| Qwen3-Omni2026.05 | 71.28 | |
| GPT-4o Audio2026.05 | 64.16 | |
| Kimi-Audio2026.05 | 62.59 | |
| MiMo-Audio2026.05 | 59.92 | |
| Qwen2.5-Omni2026.05 | 57.16 | |
| Qwen2-Audio + PCLM + DPOPCLM=true, Training=DPO2026.05 | 55.43 | |
| Qwen2-Audio + PCLMPCLM=true2026.05 | 53.3 | |
| AF32026.05 | 51.43 | |
| Qwen2-Audio2026.05 | 50.82 | |
| Qwen2-Audio + SFT w/o PCLMTraining=SFT, PCLM=false2026.05 | 50.74 | |
| AF3 + PCLM + DPOPCLM=true, Training=DPO2026.05 | 50.62 | |
| Qwen2-Audio + SFT + DPO w/o PCLMTraining=SFT + DPO, PCLM=false2026.05 | 50.36 | |
| AF3 + PCLMPCLM=true2026.05 | 50.04 | |
| AF3 + SFT w/o PCLMTraining=SFT, PCLM=false2026.05 | 49.98 | |
| VITA-Audio2026.05 | 47.71 | |
| AF3 + SFT + DPO w/o PCLMTraining=SFT + DPO, PCLM=false2026.05 | 47.1 | |
| AF22026.05 | 41.35 | |
| SALMONN2026.05 | 11.83 |