ResearchTasksMulti-modal Retrieval-Augmented GenerationMulti-modal Retrieval-Augmented GenerationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedMRAG-BenchGT (Human-Annotated Image as Input)69.99Accuracy9May 14, 2026MC-SEARCH Text-Only Chain 1.0 (test)InternVL3.5-8B + SEARCH-ALIGN38.45F18Mar 4, 2026MC-SEARCH Parallel Image-Text Fork 1.0 (test)Gemini-2.5-Pro34.83F1 Score8Mar 4, 2026MC-SEARCH Text-Initiated Chain 1.0 (test)Gemini-2.5-Pro45.3F18Mar 4, 2026MC-SEARCH Multi-Images Fork 1.0 (test)Gemini-2.5-Pro40.37F18Mar 4, 2026MC-SEARCH Image-Initiated Chain 1.0 (test)Gemini-2.5-Pro47.61F18Mar 4, 2026