Fact-based Question Answering on FVQA (test)
72.61AccuracyGemini-3-Pro
Evaluation results
| Method | Links | |
|---|---|---|
| Gemini-3-ProWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 72.61 | |
| MM-DeepResearch 32BEvaluation Paradigm=Agentic Search2026.03 | 70.1 | |
| TAPOWorkflow=Agentic Model, Model Type=Open-source, Backbone=SAPO2026.06 | 69.89 | |
| MM-DeepResearch-8BEvaluation Paradigm=Agentic Search2026.03 | 69.2 | |
| GPT-5.2Workflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 68.78 | |
| SkyWork-R1V4Workflow=Agentic Model, Model Type=Open-source2026.06 | 67.2 | |
| SenseNova-MARSWorkflow=Agentic Model, Model Type=Open-source2026.06 | 67.11 | |
| SenseNova-MARS-8BEvaluation Paradigm=Agentic Search2026.03 | 67.1 | |
| GPT-4oWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 66.34 | |
| GPT-4oEvaluation Paradigm=RAG Workflow2026.03 | 66.3 | |
| Qwen3-VL-8B-Instruct+SAPOWorkflow=Agentic Model, Model Type=Open-source2026.06 | 65.61 | |
| Gemini-3-FlashWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 64.89 | |
| GPT-5Workflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 62.61 | |
| GPT-5Evaluation Paradigm=RAG Workflow2026.03 | 62.6 | |
| MM-DeepResearch-7BEvaluation Paradigm=Agentic Search2026.03 | 61.9 | |
| Gemini-2.5-FlashWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 61.72 | |
| DeepEyes-v2-7BEvaluation Paradigm=Agentic Search2026.03 | 60.6 | |
| DeepEyesV2Workflow=Agentic Model, Model Type=Open-source2026.06 | 60.6 | |
| Qwen3-VL-32BEvaluation Paradigm=Agentic Search2026.03 | 60.2 | |
| Gemini-3-ProWorkflow=Direct Answer, Model Type=Proprietary2026.06 | 59.22 | |
| Qwen3-VL-8BEvaluation Paradigm=Agentic Search2026.03 | 58.7 | |
| MMSearch-R1-7BEvaluation Paradigm=Agentic Search2026.03 | 58.4 | |
| MMSearch-R1Workflow=Agentic Model, Model Type=Open-source2026.06 | 58.4 | |
| Gemini-3-FlashWorkflow=Direct Answer, Model Type=Proprietary2026.06 | 56.5 | |
| GPT-5Evaluation Paradigm=Direct Reasoning2026.03 | 54.4 | |
| GPT-5Workflow=Direct Answer, Model Type=Proprietary2026.06 | 54.39 | |
| Qwen3-VL-32B-InstructWorkflow=Agentic Model (zero-shot), Model Type=Open-source2026.06 | 54.28 | |
| Qwen3-VL-8B-InstructWorkflow=Agentic Model (zero-shot), Model Type=Open-source2026.06 | 53.61 | |
| Qwen3-VL-8BEvaluation Paradigm=RAG Workflow2026.03 | 53.6 | |
| Qwen2.5-VL-32B-InstructWorkflow=Agentic Model (zero-shot), Model Type=Open-source2026.06 | 52.22 | |
| GPT-5.2Workflow=Direct Answer, Model Type=Proprietary2026.06 | 50.94 | |
| GPT-4o-miniWorkflow=Agentic Model (zero-shot), Model Type=Proprietary2026.06 | 50 | |
| GPT-4oEvaluation Paradigm=Direct Reasoning2026.03 | 48 | |
| GPT-4oWorkflow=Direct Answer, Model Type=Proprietary2026.06 | 48 | |
| Gemini-2.5-FlashWorkflow=Direct Answer, Model Type=Proprietary2026.06 | 43.78 | |
| Visual-ARFTWorkflow=Agentic Model, Model Type=Open-source2026.06 | 41.72 | |
| Visual-ARFT-7BEvaluation Paradigm=Agentic Search2026.03 | 41.7 | |
| GPT-4o-miniWorkflow=Direct Answer, Model Type=Proprietary2026.06 | 36.83 | |
| Qwen2.5-VL-7B-InstructWorkflow=Agentic Model (zero-shot), Model Type=Open-source2026.06 | 36 | |
| Qwen3-VL-32BEvaluation Paradigm=Direct Reasoning2026.03 | 34.1 | |
| Qwen3-VL-32B-InstructWorkflow=Direct Answer, Model Type=Open-source2026.06 | 32.17 | |
| Qwen2.5-VL-32B-InstructWorkflow=Direct Answer, Model Type=Open-source2026.06 | 30.5 | |
| Qwen2.5-VL-7B-InstructWorkflow=Direct Answer, Model Type=Open-source2026.06 | 26.28 | |
| Qwen3-VL-8B-InstructWorkflow=Direct Answer, Model Type=Open-source2026.06 | 24.22 | |
| Qwen3-VL-8BEvaluation Paradigm=Direct Reasoning2026.03 | 24.2 |