Spoken understanding and reasoning on URO-Bench English Pro
75.4U ScoreListen-Write-Speak (w/o write while listen)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Listen-Write-Speak (w/o write while listen)visible writing during listening=disabled2026.05 | 75.4 | 84.6 | 66.8 | 75.6 | |
| Listen-Write-Speak2026.05 | 74.6 | 89 | 70.3 | 78 | |
| Listen-Write-Speak (w/o write while speak)visible writing during speaking=disabled2026.05 | 73 | 81.3 | 52.2 | 68.8 | |
| Step-Audio 22026.05 | 64.9 | 67.8 | 66.3 | 66.1 | |
| GPT-4o-Audio2026.05 | 60.7 | 64.4 | 78.5 | 67.5 | |
| GPT-Realtime2026.05 | 59.7 | 74.5 | 76.1 | 68.9 | |
| Kimi-Audio2026.05 | 50.3 | 40.6 | 56 | 49.8 | |
| Qwen-Omni2026.05 | 44.5 | 63.9 | 49.4 | 51 |