Spoken Dialogue on URO-Bench Basic Track
85.82Understanding AccuracyGLM-4-Voice
Evaluation Results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GLM-4-VoiceSize=9B2025.09 | 85.82 | — | — | — | — | — | — | — | — | — | — | — | 61.63 | 69.9 | |
| Kimi-AudioSize=7B2025.09 | 83.89 | — | — | — | — | — | — | — | — | — | — | — | 53.88 | 54.44 | |
| Pretrain+TtTSize=3B2025.09 | 57.63 | — | — | — | — | — | — | — | — | — | — | — | 39.3 | 45.68 | |
| LLaMA-OmniSize=8B2025.09 | 53.71 | — | — | — | — | — | — | — | — | — | — | — | 41.93 | 64.05 | |
| VITA-AudioSize=7B2025.09 | 52.08 | — | — | — | — | — | — | — | — | — | — | — | 51.45 | 54.97 | |
| TtTSize=3B2025.09 | 43.39 | — | — | — | — | — | — | — | — | — | — | — | 24 | 30.08 | |
| Qwen2.5-3B (AR)Size=3B2025.09 | 34.32 | — | — | — | — | — | — | — | — | — | — | — | 13.15 | 23.68 | |
| SLAM-OmniSize=0.5B2025.09 | 31.55 | — | — | — | — | — | — | — | — | — | — | — | 26.45 | 42.2 | |
| MoshiSize=7B2025.09 | 18.23 | — | — | — | — | — | — | — | — | — | — | — | 24.21 | 36.65 | |
| Mini-OmniSize=0.5B2025.09 | 15.01 | — | — | — | — | — | — | — | — | — | — | — | 14.8 | 29.71 | |
| SpeechGPTSize=7B2025.09 | 9.26 | — | — | — | — | — | — | — | — | — | — | — | 13.34 | 35.5 | |
| Qwen2.5-3B (NAR)Size=3B2025.09 | 7.22 | — | — | — | — | — | — | — | — | — | — | — | 10.12 | 20.01 | |
| ChromaLLM Scale=4B2026.01 | — | 69.05 | 74.12 | 38.61 | 71.14 | 51.69 | 22.74 | 60.26 | 60.47 | 62.07 | 64.24 | 57.44 | — | — | |
| Freeze-OmniLLM Scale=7B2026.01 | — | 70.89 | 78.87 | 26.29 | 57.74 | 46.95 | 2.81 | 42.56 | 52.23 | 48.7 | 55.8 | 48.28 | — | — | |
| GLM-4-VoiceLLM Scale=9B2026.01 | — | 90.95 | 91.07 | 64.47 | 73.8 | 59.28 | 30.93 | 57.82 | 80.77 | 63.07 | 78.76 | 69.09 | — | — | |
| LLaMA-OmniLLM Scale=8B2026.01 | — | 45.62 | 80.68 | 16.06 | 50.65 | 45.13 | 3.89 | 44.44 | 64.36 | 58.4 | 72.19 | 48.14 | — | — | |
| Mini-OmniLLM Scale=0.5B2026.01 | — | 5.07 | 32.2 | 0 | 23.25 | 25.06 | 0 | 2.82 | 30.99 | 29.8 | 31.42 | 18.06 | — | — | |
| Mini-Omni2LLM Scale=0.5B2026.01 | — | 8.1 | 40.06 | 0.66 | 28.49 | 26.92 | 0 | 6.97 | 34.81 | 30.7 | 36.43 | 21.31 | — | — | |
| SLAM-OmniLLM Scale=0.5B2026.01 | — | 12.26 | 66.21 | 1.32 | 36.95 | 34.65 | 0 | 21.85 | 48.98 | 41.03 | 52.61 | 31.59 | — | — |