Share your thoughts, 1 month free Claude Pro on usSee more

Multimodal Search on BrowseComp-VL (Accuracy)

49.1Accuracy (BrowseComp-VL)

GPT-5

Updated 1mo ago

Evaluation Results

Method	Links
GPT-5 2026.06		49.1
Visual-Seeker 2026.06		47.6
GPT-5 2026.06		47.2
Gemini-2.5-Pro 2026.06		45.2
Points-Seeker-8B 2026.06		44.4
Claude-4-Sonnet 2026.06		44.1
Gemini-2.5-Pro 2026.06		43.1
Gemini-2.5-Flash 2026.06		43.1
MM-DeepResearch-32B 2026.06		43
Vision-DeepResearch-8B 2026.06		42.6
Skywork-R1V4-30B-A3B 2026.06		38.4
MM-DeepResearch-8B 2026.06		37.9
OpenSearch-VL-8B 2026.06		37.6
Gemini-2.5-Flash 2026.06		37.1
Gemini-3-Pro 2026.03		35
Gemini-3-Flash 2026.03		31.3
VSearcher 2026.03		30.8
VSearcher-8B 2026.06		30.8
Claude-4-Sonnet 2026.06		29.3
GPT-5 2026.03		29
Qwen3-VL-8B-Instruct 2026.06		28.4
Qwen3-VL-Plus 2026.03		28
Qwen3-VL-30B-A3B-Thinking 2026.03		27.5
WebWatcher-32B 2026.06		27
Qwen3-VL-8B-Instruct 2026.06		25.1
Qwen3-VL-8B-Thinking 2026.03		24.8
Qwen2.5-VL-72B-Instruct 2026.03		24.3
DeepEyesV2 2026.03		22
Qwen3-VL-4B-Thinking 2026.03		21.5
WebWatcher-7B 2026.06		21.2
InternVL3.5-38B 2026.03		19.3
InternVL3.5-8B 2026.03		17.8
GPT-4o 2026.03		17.3
MMSearch-R1 2026.03		16