Visual Question Answering on LiveVQA
77.7AccuracyPOINTS-Seeker-8B
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| POINTS-Seeker-8BWorkflow=Agentic Search MLLM2026.04 | 77.7 | — | — | — | — | |
| Vision-DeepResearch-30B-A3BWorkflow=Agent Workflow2026.01 | 77.6 | — | — | — | — | |
| Vision-DeepResearch-8BWorkflow=Agent Workflow2026.01 | 76.7 | — | — | — | — | |
| Vision-DeepResearch-8BWorkflow=Agentic Search MLLM2026.04 | 76.7 | — | — | — | — | |
| Gemini-2.5 ProWorkflow=Agent Workflow2026.01 | 76 | — | — | — | — | |
| GPT-5Workflow=Agent Workflow2026.01 | 73.3 | — | — | — | — | |
| GPT-5Workflow=Agent Workflow2026.04 | 73.3 | — | — | — | — | |
| Gemini-2.5 FlashWorkflow=Agent Workflow2026.01 | 73 | — | — | — | — | |
| Gemini-2.5 FlashWorkflow=Agent Workflow2026.04 | 73 | — | — | — | — | |
| Claude-3.7-SonnetWorkflow=Agent Workflow2026.01 | 72 | — | — | — | — | |
| Claude-4-SonnetWorkflow=Agent Workflow2026.01 | 69.7 | — | — | — | — | |
| Claude-4-SonnetWorkflow=Agent Workflow2026.04 | 69.7 | — | — | — | — | |
| MM-DeepResearch 32BEvaluation Paradigm=Agentic Search2026.03 | 68 | — | — | — | — | |
| PIXELRAG (base)Reader=Qwen3-VL-4B, Top-k=3, Retrieval Category=Pixel-based retrieval2026.06 | 66.1 | — | — | — | 38.9 | |
| GPT-5.2Workflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 65.99 | — | — | — | — | |
| PIXELRAGReader=Qwen3-VL-4B, Top-k=3, Retrieval Category=Pixel-based retrieval2026.06 | 65.3 | — | — | — | 33.3 | |
| MM-DeepResearch-8BEvaluation Paradigm=Agentic Search2026.03 | 65 | — | — | — | — | |
| MM-DeepResearch-8BWorkflow=Agentic Search MLLM2026.04 | 65 | — | — | — | — | |
| Qwen3-VL-8B-InstructWorkflow=Agent Workflow2026.01 | 63 | — | — | — | — | |
| Qwen3-VL-30B-A3B-ThinkingWorkflow=Agent Workflow2026.01 | 62 | — | — | — | — | |
| PIXELRAG (base)Reader=Qwen3-VL-4B, top-k=1, Retrieval Modality=Pixel-based retrieval2026.06 | 61.2 | — | — | — | 26.8 | |
| PIXELRAGReader=Qwen3-VL-4B, top-k=1, Retrieval Modality=Pixel-based retrieval2026.06 | 60.8 | — | — | — | 22.4 | |
| Gemini-2.5 ProWorkflow=Direct Answer2026.01 | 60.3 | — | — | — | — | |
| Qwen3-VL-30B-A3B-InstructWorkflow=Agent Workflow2026.01 | 60 | — | — | — | — | |
| MM-DeepResearch-7BEvaluation Paradigm=Agentic Search2026.03 | 60 | — | — | — | — | |
| TAPOWorkflow=Agentic Model, Model Type=Open-source, Backbone=SAPO2026.06 | 59.71 | — | — | — | — | |
| Gemini-3-ProWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 59.69 | — | — | — | — | |
| Webwatcher-32BWorkflow=Multimodal DeepResearch MLLM2026.01 | 58.7 | — | — | — | — | |
| WebWatcher-32BEvaluation Paradigm=Agentic Search2026.03 | 58.7 | — | — | — | — | |
| WebWatcher-32BWorkflow=Agentic Search MLLM2026.04 | 58.7 | — | — | — | — | |
| GPT-5Workflow=Direct Answer2026.01 | 57.5 | — | — | — | — | |
| GPT-5Workflow=Direct Answer2026.04 | 57.5 | — | — | — | — | |
| Qwen3-VL-8B-Instruct+SAPOWorkflow=Agentic Model, Model Type=Open-source2026.06 | 56.86 | — | — | — | — | |
| Qwen3-VL-8B-ThinkingWorkflow=Agent Workflow2026.01 | 56.3 | — | — | — | — | |
| Qwen3-VL-8B-ThinkingWorkflow=Agent Workflow2026.04 | 56.3 | — | — | — | — | |
| SenseNova-MARSWorkflow=Agentic Model, Model Type=Open-source2026.06 | 56.22 | — | — | — | — | |
| SenseNova-MARS-8BEvaluation Paradigm=Agentic Search2026.03 | 56.2 | — | — | — | — | |
| SenseNova-MARS-8BWorkflow=Agentic Search MLLM2026.04 | 56.2 | — | — | — | — | |
| GPT-5Workflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 56.02 | — | — | — | — | |
| GPT-5Evaluation Paradigm=RAG Workflow2026.03 | 56 | — | — | — | — | |
| TrafilaturaReader=Qwen3-VL-4B, Top-k=3, Retrieval Category=Text-based retrieval2026.06 | 55.9 | — | — | — | 16.2 | |
| PMSRBackbone=Qwen2.5-VL-7B2025.08 | 54.2 | — | — | — | — | |
| No retrievalReader=Qwen3-VL-4B, top-k=1, Retrieval Modality=None2026.06 | 54.2 | — | — | — | — | |
| No retrievalReader=Qwen3-VL-4B, Top-k=3, Retrieval Category=None2026.06 | 54.2 | — | — | — | — | |
| TrafilaturaReader=Qwen3-VL-4B, top-k=1, Retrieval Modality=Text-based retrieval2026.06 | 53 | — | — | — | 9.8 | |
| Webwatcher-7BWorkflow=Multimodal DeepResearch MLLM2026.01 | 51.2 | — | — | — | — | |
| WebWatcher-7BEvaluation Paradigm=Agentic Search2026.03 | 51.2 | — | — | — | — | |
| WebWatcherBackbone=Qwen2.5-VL-7B2025.08 | 51.2 | — | — | — | — | |
| WebWatcher-7BWorkflow=Agentic Search MLLM2026.04 | 51.2 | — | — | — | — | |
| Gemini-2.5 FlashWorkflow=Direct Answer2026.01 | 51 | — | — | — | — | |
| Gemini-2.5 FlashWorkflow=Direct Answer2026.04 | 51 | — | — | — | — | |
| Qwen3-VL-8BEvaluation Paradigm=Agentic Search2026.03 | 50.6 | — | — | — | — | |
| Qwen3-VL-30B-A3B-ThinkingWorkflow=Direct Answer2026.01 | 49 | — | — | — | — | |
| MMSearch-R1-7BWorkflow=Multimodal DeepResearch MLLM2026.01 | 48.4 | — | — | — | — | |
| MMSearch-R1-7BEvaluation Paradigm=Agentic Search2026.03 | 48.4 | — | — | — | — | |
| MMSearch-R1Backbone=Qwen2.5-VL-7B2025.08 | 48.4 | — | — | — | — | |
| MMSearch-R1-7BWorkflow=Agentic Search MLLM2026.04 | 48.4 | — | — | — | — | |
| MMSearch-R1Workflow=Agentic Model, Model Type=Open-source2026.06 | 48.4 | — | — | — | — | |
| Gemini-3-FlashWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 48.06 | — | — | — | — | |
| Gemini-2.5-FlashWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 47.75 | — | — | — | — | |
| GPT-5.2Workflow=Direct Answer, Model Type=Proprietary2026.06 | 47 | — | — | — | — | |
| Qwen3-VL-32BEvaluation Paradigm=Direct Reasoning2026.03 | 45.5 | — | — | — | — | |
| Qwen3-VL-32BEvaluation Paradigm=Agentic Search2026.03 | 45.5 | — | — | — | — | |
| GPT-5Evaluation Paradigm=Direct Reasoning2026.03 | 44.4 | — | — | — | — | |
| GPT-5Workflow=Direct Answer, Model Type=Proprietary2026.06 | 44.39 | — | — | — | — | |
| GoG-3-8B-Think-RLEvaluation Methodology=Search-Equipped Models2026.01 | 43.85 | — | — | — | — | |
| Qwen3-VL-8B-ThinkingWorkflow=Direct Answer2026.01 | 43.3 | — | — | — | — | |
| MIAAgent Type=Memory-based Search Agent2026.04 | 43.1 | — | — | — | — | |
| Qwen3-VL-32B-InstructWorkflow=Agentic Model (zero-shot), Model Type=Open-source2026.06 | 42.87 | — | — | — | — | |
| Qwen3-VL-30B-A3B-InstructWorkflow=Direct Answer2026.01 | 42.7 | — | — | — | — | |
| Qwen2.5-VL-32B-InstructWorkflow=Agentic Model (zero-shot), Model Type=Open-source2026.06 | 42.17 | — | — | — | — | |
| Gemini-2.5-flashWorkflow=RAG Workflow2026.01 | 41.3 | — | — | — | — | |
| Gemini-2.5 FlashWorkflow=RAG Workflow2026.04 | 41.3 | — | — | — | — | |
| Qwen3-VL-8B-InstructWorkflow=Direct Answer2026.01 | 41 | — | — | — | — | |
| Qwen3-VL-8B-InstructWorkflow=Direct Answer2026.04 | 41 | — | — | — | — | |
| OmniSearchBackbone=GPT-4o2025.08 | 40.9 | — | — | — | — | |
| GPT-4oEvaluation Paradigm=RAG Workflow2026.03 | 40.1 | — | — | — | — | |
| Unsupervised MIAAgent Type=Memory-based Search Agent2026.04 | 40.1 | — | — | — | — | |
| GPT-4oWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 40.09 | — | — | — | — | |
| Gemini-3-ProWorkflow=Direct Answer, Model Type=Proprietary2026.06 | 40.06 | — | — | — | — | |
| Qwen3-VL-8B-InstructWorkflow=Agentic Model (zero-shot), Model Type=Open-source2026.06 | 39.37 | — | — | — | — | |
| Qwen3-VL-8BEvaluation Paradigm=RAG Workflow2026.03 | 39.3 | — | — | — | — | |
| Gemini-3-FlashWorkflow=Direct Answer, Model Type=Proprietary2026.06 | 38.9 | — | — | — | — | |
| Claude-4-SonnetWorkflow=Direct Answer2026.01 | 38.5 | — | — | — | — | |
| Claude-4-SonnetWorkflow=Direct Answer2026.04 | 38.5 | — | — | — | — | |
| Claude-3.7-SonnetWorkflow=Direct Answer2026.01 | 38 | — | — | — | — | |
| GoG-2.5-7B-RLEvaluation Methodology=Search-Equipped Models2026.01 | 37.7 | — | — | — | — | |
| DR-MMSearchAgentEvaluation Paradigm=Search Agent2026.04 | 37 | — | — | — | — | |
| MementoAgent Type=Memory-based Search Agent2026.04 | 36.7 | — | — | — | — | |
| Qwen-2.5-VL-72BWorkflow=RAG Workflow2026.01 | 35.7 | — | — | — | — | |
| Qwen-2.5-VL-72BWorkflow=RAG Workflow2026.04 | 35.7 | — | — | — | — | |
| ReasoningBankAgent Type=Memory-based Search Agent2026.04 | 34.2 | — | — | — | — | |
| ExpeLAgent Type=Memory-based Search Agent2026.04 | 34.1 | — | — | — | — | |
| GPT-4oWorkflow=RAG Workflow2026.04 | 34 | — | — | — | — | |
| No MemoryAgent Type=Memory-based Search Agent2026.04 | 33 | — | — | — | — | |
| GoG-3-8B-Think-SFTEvaluation Methodology=Search-Equipped Models2026.01 | 32.4 | — | — | — | — | |
| RAGAgent Type=Memory-based Search Agent2026.04 | 31.7 | — | — | — | — | |
| Qwen3-VL-32B-InstructWorkflow=Direct Answer, Model Type=Open-source2026.06 | 31.59 | — | — | — | — | |
| Gemini-2.5-FlashWorkflow=Direct Answer, Model Type=Proprietary2026.06 | 31.57 | — | — | — | — | |
| GPT-4o-miniWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 31.54 | — | — | — | — |