Audio Reasoning (Multiple Choice) on SpeechR MC
70.5AccuracyStep-Audio-R1
Evaluation Results
| Method | Links | |
|---|---|---|
| Step-Audio-R12025.08 | 70.5 | |
| VIRBABackbone=Qwen2.5-Omni2025.08 | 67.4 | |
| AF3Full name=Audio Flamingo 32025.08 | 64.2 | |
| VIRBABackbone=Qwen2-Audio2025.08 | 62.8 | |
| Kimi-Audio2025.08 | 61.8 | |
| Qwen2.5-Omni2025.08 | 57.4 | |
| Single-view RL2025.08 | 52.7 | |
| GLM-4-Voice2025.08 | 51 | |
| Qwen2-Audio base2025.08 | 45.8 |