Response type prediction on Full-Videos (test) benchmark leaderboard · SOTA2 Research

Response type prediction on Full-Videos (test)

31.55Affm. PrecisionMM-When2Speak

Evaluation Results

Method	Links
MM-When2Speak2025.05		31.55	29.24	32.26	29.95	31.25	28.94	29.53	27.22	28.21	25.91	28.82	26.52	27.47	25.17	35.17	32.85	33.27	30.95
VITA-1.52025.05		21.4	18.95	18.94	16.51	22.18	19.72	17.73	15.31	21.97	19.52	22.15	19.67	20.72	18.27	26.21	23.73	27.05	24.57
VITA-1.52025.05		20.06	17.27	17.44	14.68	18.76	15.98	17.48	14.72	18.51	15.73	18.27	15.5	18.94	16.16	21.24	18.44	21.15	18.35
ChatGPT-5.22025.05		19.95	17	18.01	15.08	20.9	17.94	20.44	17.49	19.58	16.64	21.39	18.43	19.83	16.88	22.46	19.49	21.45	18.5
Gemini-3.12025.05		19.77	16.93	18.01	15.19	18.41	15.38	18.25	15.27	20.83	17.98	17.86	15.05	19.25	16.41	19.94	17.09	19.56	16.73
Qwen2.5-VL2025.05		15.44	18.58	17.67	16.03	15.54	17.12	18.03	14.52	13.7	13.95	15.25	17.25	13.64	17.51	18.44	15.12	18.38	16.88
ChatGPT-5.22025.05		14.15	11.64	12.5	10.02	15.54	12.99	13.86	11.37	16.31	13.68	14.42	11.9	16.19	13.65	15.38	12.85	13.22	10.73
Gemini-3.12025.05		13.9	11.49	13.38	10.97	14.25	11.78	13.62	11.2	14.43	12.01	14.15	11.72	15.18	12.74	13.22	10.81	13.27	10.86
VITA-1.52025.05		12.97	10.61	12.62	10.26	12.75	10.39	11.53	9.19	15.09	12.7	15.33	12.93	14.07	11.69	15.33	12.93	13.6	11.23
Qwen2-Audio2025.05		11.99	9.64	12.77	10.41	13.29	10.92	13.86	11.48	15.19	12.79	14.57	12.18	13.21	10.84	14.26	11.88	11.8	9.46
ChatGPT-5.22025.05		11.5	8.71	9.36	6.65	8.58	5.89	8.66	5.98	8.53	5.85	9.32	6.6	8.89	6.2	11.64	8.84	12.22	9.41
Qwen-2.52025.05		11.29	8.64	7.55	5.05	9.78	7.18	8.92	6.35	9.17	6.59	9.59	7	7.95	5.43	10.47	7.85	10.69	8.06
VITA-1.52025.05		10.38	7.76	9.64	7.04	9.47	6.88	7.95	5.43	8.52	5.97	8.39	5.85	9.25	6.67	10.53	7.9	11.18	8.53
Gemini-3.12025.05		10.19	7.53	8.53	5.94	8.74	6.14	8.23	5.65	8.7	6.1	7.84	5.28	8.68	6.08	9.56	6.92	11.37	8.68
DeepSeek-V32025.05		9.46	6.87	7.69	5.18	6.82	4.36	8.34	5.8	7.69	5.18	7.97	5.45	8.72	6.16	10.39	7.77	10.95	8.31