Paralinguistic Perception on MMSU Paralinguistic
54.51Para. ScoreStep-Audio-R1
Evaluation Results
| Method | Links | |
|---|---|---|
| Step-Audio-R1Model Type=Open-Source Audio LLMs2026.05 | 54.51 | |
| Gemini 2.5 FlashModel Type=Closed-Source Models2026.05 | 51.05 | |
| Qwen3-OmniModel Type=Open-Source Omni LLMs2026.05 | 49.13 | |
| Kimi-Audio-7B-InstructModel Type=Open-Source Audio LLMs2026.05 | 41.48 | |
| Audio Flamingo 3 (AF3)Model Type=Open-Source Audio LLMs2026.05 | 37.74 | |
| GPT-4o AudioModel Type=Closed-Source Models2026.05 | 36.55 | |
| MiMo-Audio-7B-InstructModel Type=Open-Source Audio LLMs2026.05 | 35.64 | |
| Qwen2-Audio-7B-InstructModel Type=Open-Source Audio LLMs2026.05 | 34.37 | |
| Qwen2.5-Omni-7BModel Type=Open-Source Omni LLMs2026.05 | 33.45 | |
| VITA-AudioModel Type=Open-Source Audio LLMs2026.05 | 29.54 | |
| Audio Flamingo 2 (AF2)Model Type=Open-Source Audio LLMs2026.05 | 27.44 | |
| SALMONN-7BModel Type=Open-Source Audio LLMs2026.05 | 6.84 |