Omnimodal Question Answering on OmniVideoBench 1.0 (test) benchmark leaderboard · SOTA2 Research

Omnimodal Question Answering on OmniVideoBench 1.0 (test)

44.44Compare AttrGemini 2.0-Flash + RAG

Evaluation Results

Method	Links
Gemini 2.0-Flash + RAG2026.02		44.44	36.36	12.5	43.48	28.17	24.59	20	46.88	29.69
Gemini 2.5-Flash + RAG + Agent2026.02		44.44	27.27	29.17	30.43	33.8	34.43	28	53.12	35.16
Qwen2.5-Omni-7B + RAG + Agent2026.02		44.44	45.45	29.17	26.09	28.17	22.95	24	34.52	28.52
Qwen2.5-Omni-3B + RAG2026.02		33.33	27.27	33.33	26.09	26.76	16.39	24	25	24.61
Qwen2.5-Omni-3B + RAG + Agent + RL2026.02		33.33	27.27	33.33	21.74	26.76	21.31	36	31.25	27.34
Qwen2.5-Omni-7B + RAG + Agent + RL2026.02		33.33	27.27	12.5	26.09	32.39	34.43	20	37.5	29.69
Qwen3-Omni-30B2026.02		33.33	18.18	12.5	26.09	32.39	32.79	12	34.38	27.73
Qwen3-Omni-30B + RAG + Agent2026.02		33.33	36.36	16.67	21.74	38.03	27.87	28	25	29.3
GPT-5.12026.02		22.22	27.27	20.83	26.09	29.17	19.67	4	40.62	24.51
Gemini 2.0-Flash2026.02		22.22	36.36	25	30.43	26.76	29.51	16	31.25	27.34
Gemini 2.0-Flash + RAG + Agent2026.02		22.22	36.36	16.67	34.78	29.58	29.51	24	46.88	30.47
Gemini 2.5-Flash2026.02		22.22	18.18	25	34.78	35.21	31.15	4	31.25	28.52
Qwen2.5-Omni-3B2026.02		22.22	9.09	25	17.39	25.35	18.03	28	31.25	23.05
Qwen2.5-Omni-3B + RAG + Agent2026.02		22.22	26.36	37.5	26.09	25.35	24.59	24	28.12	26.95
Qwen2.5-Omni-7B2026.02		22.22	18.18	25	26.09	23.94	27.87	16	31.25	25
Qwen2.5-Omni-7B + RAG2026.02		22.22	27.27	12.5	39.13	22.54	24.59	40	40.62	27.73
Gemini 2.5-Flash + RAG2026.02		11.11	36.36	20.83	30.43	33.8	36.07	12	53.12	32.42
Qwen3-Omni-30B + RAG2026.02		11.11	36.36	8.33	26.09	38.03	26.23	20	34.38	28.12