Voice Evaluation on VoiceBench
90.1SD Score (VoiceBench)Gemini 2.5 Pro (May 25)
Evaluation Results
| Method | Links | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Gemini 2.5 Pro (May 25)Model Category=Closed-Source APIs2026.04 | 90.1 | 89.6 | 94.3 | 88.4 | 93.4 | 71.1 | 92.3 | 92.6 | 85.7 | 98.1 | — | — | — | — | — | |
| Ultravox-GLM-4P72026.07 | 84.2 | — | — | — | — | — | — | 87.2 | 76.3 | 99.2 | 97.4 | 86 | 91 | 83.8 | 94.7 | |
| Qwen3-Omni 30B-A3Bvariant=Thinking2026.07 | 78.1 | — | — | — | — | — | — | 88.9 | 80.6 | 97.2 | 96.4 | 90.5 | 90.5 | 83 | 94.3 | |
| Qwen3-Omni (Sep 25)Model Category=Omnimodal MLLMs2026.04 | 76.9 | 85.5 | 94.8 | 90.8 | 91.6 | 68.1 | 89.7 | 80.4 | 77.8 | 99.3 | — | — | — | — | — | |
| Qwen3-Omni 30B-A3Bvariant=Instruct2026.07 | 76.9 | — | — | — | — | — | — | 80.4 | 77.8 | 99.3 | 94.8 | 90.8 | 91.6 | 68.1 | 89.7 | |
| GPT-4o (May 24)Model Category=Closed-Source APIs2026.04 | 75.5 | 86.8 | 95.6 | 89.8 | 91.6 | 80.3 | 89.2 | 84.1 | 76 | 98.7 | — | — | — | — | — | |
| GPT-4o-Audio2026.07 | 75.5 | — | — | — | — | — | — | 84.1 | 76 | 98.7 | 95.6 | 89.8 | 91.6 | 80.2 | 89.2 | |
| Kimi-Audio-Instruct2026.07 | 63.1 | — | — | — | — | — | — | 69.7 | 61.1 | 100 | 89.2 | 79.4 | 84 | 62.2 | 83.5 | |
| Audexsize=30B-A3B2026.07 | 60.2 | — | — | — | — | — | — | 88 | 88 | 99.6 | 87.4 | 82.7 | 78.2 | 82.6 | 91 | |
| DeSTA 2.5 (Jul 25)Model Category=Audio MLLMs2026.04 | 60 | 74.5 | 88.2 | 79.8 | 76 | 60.9 | 74.1 | 67 | 66.4 | 98.3 | — | — | — | — | — | |
| Qwen2.5-Omni (Mar 25)Model Category=Omnimodal MLLMs2026.04 | 56.4 | 73.6 | 89.9 | 76.7 | 77.7 | 61.7 | 80.9 | 66.7 | 53.5 | 99.2 | — | — | — | — | — | |
| AudioReaonser (Sep 25)Model Category=Audio MLLMs2026.04 | 47.5 | 51.7 | 73.4 | 67.5 | 60.5 | 35.6 | 53.6 | 54 | 24.7 | 48.7 | — | — | — | — | — | |
| Omni-o3Model Category=Omnimodal MLLMs2026.04 | 46.4 | 57.7 | 85 | 49.2 | 66.6 | 31.7 | 33.6 | 63.5 | 47 | 96 | — | — | — | — | — | |
| Baichuan-Audio2026.07 | 45.8 | — | — | — | — | — | — | 54.8 | 50.3 | 99.4 | 88.2 | 81.6 | 78.4 | 53.2 | 71.7 | |
| Phi-4-multimodal2026.07 | 39.8 | — | — | — | — | — | — | 61.8 | 45.4 | 100 | 76.2 | 76.4 | 71.2 | 42.2 | 65.9 | |
| MiMo-Audio2026.07 | 36.3 | — | — | — | — | — | — | 68.4 | 69.5 | 98.3 | 75 | 64.6 | 63.5 | 52.8 | 66.2 | |
| AFlamingo3 (NeurIPS 25)Model Category=Audio MLLMs2026.04 | 36 | 38.1 | 42.6 | 35.8 | 37.1 | 41 | 57.4 | 52.1 | 29.2 | 11.7 | — | — | — | — | — | |
| Qwen2-Audio (Jul 24)Model Category=Audio MLLMs2026.04 | 35.7 | 55.8 | 74.8 | 68.6 | 60.2 | 35.7 | 49.5 | 54.7 | 26.3 | 96.7 | — | — | — | — | — | |
| Ola (Feb 25)Model Category=Omnimodal MLLMs2026.04 | 33.8 | 59.4 | 82.4 | 59.4 | 63.8 | 46 | 67.9 | 51.1 | 39.6 | 90.8 | — | — | — | — | — | |
| Audexsize=2B2026.07 | 25.7 | — | — | — | — | — | — | 64.4 | 73.8 | 99.8 | 70.1 | 66.9 | 60.5 | 62.1 | 73 |