Speech Understanding and Fluency on VocalBench
91.3Knowledge ScoreGPT-4o
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| GPT-4o2026.06 | 91.3 | 4.69 | 3.93 | 4.16 | 4.67 | 88.06 | |
| Qwen3-Omni-30B2026.06 | 89.4 | 4.43 | 4.77 | 4.38 | 4.96 | 91.99 | |
| Qwen2.5-Omni-7B2026.06 | 69.5 | 4.36 | 3.18 | 4.17 | 3.54 | 74.92 | |
| PRIME-Speech-9B2026.06 | 68.9 | 4.23 | 3.37 | 4.36 | 4.29 | 78.76 | |
| VocalNet-8Bstreaming mode=true2026.06 | 68 | 3.75 | 3.51 | 4.45 | 3.53 | 74.54 | |
| Kimi-Audio-7B2026.06 | 62.2 | 3.13 | 3.1 | 2.36 | 3.15 | 59.38 | |
| Step-Audio-2-Mini-7B2026.06 | 58.5 | 3.67 | 3.12 | 4.52 | 3.44 | 70.71 | |
| GLM-4-Voice-9B2026.06 | 56.4 | 3.64 | 3.29 | 3.87 | 3.62 | 68.94 |