Visual Question Answering on BC-VL
53.7AccuracyVision-DeepResearch-30B-A3B
Evaluation Results
| Method | Links | |
|---|---|---|
| Vision-DeepResearch-30B-A3BWorkflow=Agent Workflow2026.01 | 53.7 | |
| Claude-3.7-SonnetWorkflow=Agent Workflow2026.01 | 50.4 | |
| Gemini-2.5 ProWorkflow=Agent Workflow2026.01 | 49.9 | |
| Claude-4-SonnetWorkflow=Agent Workflow2026.01 | 48.6 | |
| GPT-5Workflow=Direct Answer2026.01 | 47.2 | |
| GPT-5Workflow=Agent Workflow2026.01 | 46.1 | |
| Gemini-2.5 FlashWorkflow=Agent Workflow2026.01 | 44.6 | |
| Qwen3-VL-30B-A3B-ThinkingWorkflow=Agent Workflow2026.01 | 44.1 | |
| Gemini-2.5 ProWorkflow=Direct Answer2026.01 | 43.1 | |
| Vision-DeepResearch-8BWorkflow=Agent Workflow2026.01 | 42.6 | |
| Qwen3-VL-30B-A3B-InstructWorkflow=Agent Workflow2026.01 | 42.6 | |
| Qwen3-VL-8B-InstructWorkflow=Agent Workflow2026.01 | 38.6 | |
| Gemini-2.5 FlashWorkflow=Direct Answer2026.01 | 37.1 | |
| Qwen3-VL-8B-ThinkingWorkflow=Agent Workflow2026.01 | 37.1 | |
| Qwen3-VL-30B-A3B-ThinkingWorkflow=Direct Answer2026.01 | 34.6 | |
| Claude-3.7-SonnetWorkflow=Direct Answer2026.01 | 32.3 | |
| Qwen3-VL-30B-A3B-InstructWorkflow=Direct Answer2026.01 | 29.6 | |
| Claude-4-SonnetWorkflow=Direct Answer2026.01 | 29.3 | |
| Webwatcher-32BWorkflow=Multimodal DeepResearch MLLM2026.01 | 26.7 | |
| Qwen3-VL-8B-InstructWorkflow=Direct Answer2026.01 | 25.1 | |
| Qwen3-VL-8B-ThinkingWorkflow=Direct Answer2026.01 | 25.1 | |
| Webwatcher-7BWorkflow=Multimodal DeepResearch MLLM2026.01 | 20.3 | |
| Gemini-2.5-flashWorkflow=RAG Workflow2026.01 | 12.1 | |
| Qwen-2.5-VL-72BWorkflow=RAG Workflow2026.01 | 10.2 | |
| Claude-3.7-SonnetWorkflow=RAG Workflow2026.01 | 10 |