Multimodal Search on MM-BrowseComp (test)
16.1AccuracyVisual-Seeker
Evaluation Results
| Method | Links | |
|---|---|---|
| Visual-SeekerWorkflow Protocol=Multimodal Deep Search Agent2026.06 | 16.1 | |
| Gemini-2.5-ProWorkflow Protocol=Agentic Workflow2026.06 | 14.8 | |
| Gemini-2.5-FlashWorkflow Protocol=Agentic Workflow2026.06 | 10.4 | |
| GPT-5Workflow Protocol=Direct Answer2026.06 | 10.3 | |
| Gemini-2.5-ProWorkflow Protocol=Direct Answer2026.06 | 10.3 | |
| GPT-5Workflow Protocol=Agentic Workflow2026.06 | 9.4 | |
| Qwen3-VL-8B-InstructWorkflow Protocol=Agentic Workflow2026.06 | 6.7 | |
| Claude-4-SonnetWorkflow Protocol=Agentic Workflow2026.06 | 6.3 | |
| VSearcher-8BWorkflow Protocol=Multimodal Deep Search Agent2026.06 | 6.2 | |
| Gemini-2.5-FlashWorkflow Protocol=Direct Answer2026.06 | 5.4 | |
| Claude-4-SonnetWorkflow Protocol=Direct Answer2026.06 | 5.3 | |
| Qwen3-VL-8B-InstructWorkflow Protocol=Direct Answer2026.06 | 4.9 |