Video Search Question Answering on VSQA
65.11ScoreGemini-3-Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3-ProWorkflow=Agentic Model, Zero-shot=true2026.07 | 65.11 | |
| Gemini-3-FlashWorkflow=Agentic Model, Zero-shot=true2026.07 | 58.99 | |
| GPT-5.2Workflow=Agentic Model, Zero-shot=true2026.07 | 57.19 | |
| Gemini-3-FlashWorkflow=Direct Answer, Zero-shot=false2026.07 | 56.12 | |
| Gemini-3-ProWorkflow=Direct Answer, Zero-shot=false2026.07 | 54.32 | |
| VideoSearcher-8BWorkflow=Agentic Model, Zero-shot=false2026.07 | 51.8 | |
| GPT-4oWorkflow=Agentic Model, Zero-shot=true2026.07 | 49.64 | |
| VideoSearcher-4BWorkflow=Agentic Model, Zero-shot=false2026.07 | 49.28 | |
| GPT-5.2Workflow=Direct Answer, Zero-shot=false2026.07 | 38.85 | |
| Qwen3-VL-32B-InstructWorkflow=Agentic Model, Zero-shot=true2026.07 | 37.77 | |
| GPT-4oWorkflow=Direct Answer, Zero-shot=false2026.07 | 36.33 | |
| Webwatcher-7BWorkflow=Agentic Model, Zero-shot=false2026.07 | 34.53 | |
| Qwen3-VL-8B-InstructWorkflow=Agentic Model, Zero-shot=true2026.07 | 30.94 | |
| Qwen3-VL-4B-InstructWorkflow=Agentic Model, Zero-shot=true2026.07 | 29.14 | |
| Qwen2.5-VL-32B-InstructWorkflow=Agentic Model, Zero-shot=true2026.07 | 29.14 | |
| DeepEyesV2Workflow=Agentic Model, Zero-shot=false2026.07 | 21.58 | |
| Qwen3-VL-32B-InstructWorkflow=Direct Answer, Zero-shot=false2026.07 | 20.86 | |
| Qwen3-VL-8B-InstructWorkflow=Direct Answer, Zero-shot=false2026.07 | 16.91 | |
| Qwen2.5-VL-32B-InstructWorkflow=Direct Answer, Zero-shot=false2026.07 | 13.67 | |
| Qwen3-VL-4B-InstructWorkflow=Direct Answer, Zero-shot=false2026.07 | 12.59 | |
| MMSearch-R1Workflow=Agentic Model, Zero-shot=false2026.07 | 11.87 | |
| Qwen2.5-VL-7B-InstructWorkflow=Agentic Model, Zero-shot=true2026.07 | 11.15 | |
| Qwen2.5-VL-7B-InstructWorkflow=Direct Answer, Zero-shot=false2026.07 | 9.35 | |
| Visual-ARFTWorkflow=Agentic Model, Zero-shot=false2026.07 | 8.99 |