Multi-modal Retrieval-Augmented Generation on MC-SEARCH Multi-Images Fork 1.0 (test) benchmark leaderboard · SOTA2 Research

Multi-modal Retrieval-Augmented Generation on MC-SEARCH Multi-Images Fork 1.0 (test)

40.37F1Gemini-2.5-Pro

Evaluation Results

Method	Links
Gemini-2.5-Pro2026.03		40.37	36.58	2.76	18.68	1.4	61.29
Qwen2.5-VL-7B + SEARCH-ALIGN2026.03		38.01	19.41	2.11	38.01	1.16	57.04
Gemini-2.5-Flash2026.03		36.8	31.89	2.35	13.45	1.66	64.4
InternVL3.5-8B + SEARCH-ALIGN2026.03		36.53	24.43	2.49	39.33	1.67	57.65
InternVL3.5-8B2026.03		33.08	20.98	2.13	16.02	2.25	57.65
Claude-3.7-Sonnet2026.03		29.78	27.85	2.07	19.43	1.45	62.41
Qwen2.5-VL-7B2026.03		26.13	7.53	1.41	12.58	3.8	57.04
GPT-4o-Mini2026.03		24	19.16	2.13	16.04	1.94	59.46