Multi-modal Retrieval-Augmented Generation on MC-SEARCH Text-Initiated Chain 1.0 (test) benchmark leaderboard · SOTA2 Research

Multi-modal Retrieval-Augmented Generation on MC-SEARCH Text-Initiated Chain 1.0 (test)

45.3F1Gemini-2.5-Pro

Evaluation Results

Method	Links
Gemini-2.5-Pro2026.03		45.3	29.89	3.62	19.51	0.95	55.94
Qwen2.5-VL-7B + SEARCH-ALIGN2026.03		45.07	27.94	2.42	33.59	0.87	39.86
Gemini-2.5-Flash2026.03		43.55	26.34	3.3	25.2	1.2	66.27
InternVL3.5-8B + SEARCH-ALIGN2026.03		40.87	28.01	2.59	37.82	1.44	38.7
InternVL3.5-8B2026.03		38.34	25.48	2.6	19.06	1.97	38.7
Claude-3.7-Sonnet2026.03		36.1	17.36	2.63	25.09	1.12	57.13
Qwen2.5-VL-7B2026.03		31.47	14.34	1.66	17.02	3.8	39.86
GPT-4o-Mini2026.03		30.11	18.46	2.41	27.18	1.76	49.47