Video Reasoning on MOV-Bench benchmark leaderboard · SOTA2 Research

Video Reasoning on MOV-Bench

71.43Causal AccuracyAOP-Agent

Evaluation Results

Method	Links
AOP-Agent2026.05		71.43	58.39	50	62.99	70	60.85	62.41	64.55	64.75	61.59	56.98	62.62
AOP-Agent2026.05		70.33	55.28	48.33	68.5	67.5	55.03	63.12	68.78	60.68	67.39	59.3	62.24
Qwen3-Omni-Captioner2026.05		59.34	57.76	30	51.97	57.5	46.56	53.19	60.32	52.88	54.35	53.49	53.37
Qwen3-Omni-Thinking2026.05		56.04	57.14	35	60.63	58.75	48.15	51.77	65.61	57.29	58.7	44.19	55.49
MiniCPM-o 4.52026.05		54.95	58.39	40	54.33	48.75	43.39	53.9	62.43	56.27	50	47.67	53.18
AOP-Agent2026.05		50.55	47.83	41.67	51.18	38.75	41.27	46.81	52.91	48.14	47.1	43.02	47.01
Qwen3-Omni-Instruct2026.05		48.35	59.63	35	54.33	55	45.5	49.65	62.43	54.92	49.28	51.16	52.79
Ming-Lite-Omni-1.52026.05		46.15	55.9	35	50.39	50	44.44	47.52	56.08	49.83	53.62	41.86	49.52
OmniAgent2026.05		45.05	38.51	28.33	35.43	42.5	33.86	39.72	41.8	36.61	45.65	32.56	38.34
Qwen2.5-Omni-3B2026.05		42.86	45.96	38.33	44.09	43.75	35.98	43.97	51.32	44.41	46.38	37.21	43.74
Qwen2.5-Omni-7B2026.05		41.76	52.8	25	55.12	42.5	37.57	45.39	56.61	49.15	45.65	39.53	46.63
Baichuan-Omni-1.52026.05		41.76	47.83	30	44.09	47.5	37.57	39.72	52.91	45.76	45.65	33.72	43.74
Active Video Perception2026.05		38.46	32.3	34.48	30.16	46.25	32.8	38.85	35.11	33.79	39.42	33.72	35.27