SOTA Multimodal Search-based Question Answering on MMSearch and PapersWithCode

69.6Accuracy

Vision-DeepResearch-8B

Updated 25d ago

Evaluation Results

Method	Links
Vision-DeepResearch-8B 2026.01		69.6
Vision-DeepResearch-30B-A3B 2026.01		69.6
Gemini-2.5 Pro 2026.01		69
Claude-4-Sonnet 2026.01		67.2
GoG-3-8B-Think-RL 2026.01		65.5
Gemini-2.5 Flash 2026.01		64
Qwen3-VL-8B-Thinking 2026.01		63.84
GPT-5 2026.01		63.7
Claude-3.7-Sonnet 2026.01		63.7
DeepEyesV2 2025.11		63.7
DeepEyesV2 2025.08		63.7
GoG-2.5-7B-RL 2026.01		56.73
Webwatcher-32B 2026.01		55.3
Qwen3-VL-30B-A3B-Instruct 2026.01		55
PMSR 2025.08		54.3
MMSearch-R1-7B 2026.01		53.8
GoG-3-8B-Think-SFT 2026.01		53.8
MMSearch-R1 2025.11		53.8
MMSearch-R1 2025.08		53.8
Qwen3-VL-30B-A3B-Thinking 2026.01		53.2
Qwen-2.5-VL 2025.11		52.2
Qwen3-VL-8B-Instruct 2026.01		52
OmniSearch 2025.08		49.7
Webwatcher-7B 2026.01		49.1
WebWatcher 2025.11		49.1
WebWatcher 2025.08		49.1
Qwen3-VL-8B-Thinking 2026.01		45.6
Gemini-2.5-flash 2026.01		43.9
MMSearch-R1* 2026.01		40.94
GoG-2.5-7B-SFT 2026.01		40.35
Gemini-2.5 Pro 2026.01		39.8
Qwen3-VL-8B-Think 2026.01		36.84
Qwen2.5-VL-7B 2026.01		36.26
GPT-5 2026.01		33.3
Claude-3.7-Sonnet 2026.01		32.7
Gemini-2.5 Flash 2026.01		30.4
Qwen-2.5-VL-72B 2026.01		29.2
Gemini 2.5 Pro 2025.11		26.9
GPT4o 2025.11		22.2
GPT-4o 2026.01		21.64
PIXELRAG 2026.06		21.6
Claude-3.7-Sonnet 2026.01		21.1
Qwen2.5-VL-7B 2026.01		20.47
PIXELRAG (base) 2026.06		19.9
Qwen3-VL-30B-A3B-Thinking 2026.01		19.3
Claude-4-Sonnet 2026.01		18.7
Qwen3-VL-30B-A3B-Instruct 2026.01		18.7
Qwen3-VL-30B-A3B-Instruct 2026.01		18.13
Qwen3-VL-30B-A3B-Thinking 2026.01		17.54
Qwen2.5-VL-32B 2026.01		16.37
Qwen3-VL-8B-Thinking 2026.01		15.8
Qwen3-VL-8B-Instruct 2026.01		15.2
Qwen3-VL-8B-Instruct 2026.01		15.2
Qwen3-VL-8B-Thinking 2026.01		15.2
Qwen2.5-VL-7B 2026.01		14.62
No retrieval 2026.06		13.5
Trafilatura 2026.06		13.5
mwparserfromhell 2026.06		12.9
Qwen-2.5-VL 2025.11		12.8