Spoken Dialogue Evaluation on VCB Bench benchmark leaderboard · SOTA2 Research

Spoken Dialogue Evaluation on VCB Bench

93.07TIFCovo-Audio-Chat

Evaluation Results

Method	Links
Covo-Audio-Chat2026.02		93.07	89.94	67.41	60.86	87.7	49.95	79.34	64.95	88.94	87.13	90.37
Qwen3-Omni2026.02		90.45	85.17	70.73	65.57	87.17	66.86	81.9	82.78	87.91	85.63	86.03
MiMo-Audio2026.02		90.08	81.89	56.26	42.74	86.3	48.7	81.75	—	83.72	85.36	89.38
Fun-Audio-Chat2026.02		89.3	89.39	78.82	75.7	85.27	46.3	73	79.15	88.6	83.83	85.15
GPT-4o Audio2026.02		86.94	88.8	77.98	82.9	33.59	55.81	73.45	76.74	80.34	79.92	86.51
Baichuan-Audio2026.02		82.24	78.6	53.67	48.27	80.96	40.73	74.05	51.66	78.05	77.62	78.71
GLM-4-Voice2026.02		82.15	75.52	73.18	66.94	82.56	41.79	60.18	—	73.64	77.51	78.6
Kimi-Audio2026.02		77.33	60.37	71.04	54.13	82.27	54.47	75.42	51.96	64.35	65.14	71.51
Qwen2.5-Omni2026.02		76.75	60.83	54.04	42.18	85.23	48.13	73	72.81	77.82	74.41	78.27
Step-Audio 2 mini2026.02		76.1	70.47	60.8	49.85	82.31	49.86	72.85	79.46	75.19	70.58	74.78