List your company
ResearchFormerly WizWand

Visual Document Reasoning on VDR

27.2AccuracyClaude-3.7-Sonnet
0.9927.79614.621.404May 11, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Claude-3.7-SonnetEvaluation Setting=Agent Workflow2026.0527.2
Qwen3-VL-30B + ODE-RLEvaluation Setting=Visual-Native Agent Harness, Training Strategy=ODE-RL2026.0526.4
Qwen3-VL-30B + ODE-SFTEvaluation Setting=Visual-Native Agent Harness, Training Strategy=ODE-SFT2026.0524
Qwen3-VL-8B + ODE-RLEvaluation Setting=Visual-Native Agent Harness, Training Strategy=ODE-RL2026.0520.4
Qwen3-VL-8B + ODE-SFTEvaluation Setting=Visual-Native Agent Harness, Training Strategy=ODE-SFT2026.0519.2
GPT-5Evaluation Setting=Agent Workflow2026.0517.6
Claude-4-SonnetEvaluation Setting=Agent Workflow2026.0513.6
Qwen3-VL-30BEvaluation Setting=Visual-Native Agent Harness2026.0511
GPT-5Evaluation Setting=Direct Reasoning2026.0510.8
Gemini-2.5 ProEvaluation Setting=Agent Workflow2026.0510
Gemini-2.5 ProEvaluation Setting=Direct Reasoning2026.058
Gemini-2.5 FlashEvaluation Setting=Agent Workflow2026.057.8
Gemini-2.5 FlashEvaluation Setting=Direct Reasoning2026.056.2
Qwen3-VL-8BEvaluation Setting=Agent Workflow2026.055
Claude-3.7-SonnetEvaluation Setting=Direct Reasoning2026.054.6
Qwen3-VL-30BEvaluation Setting=Agent Workflow2026.054.4
Qwen3-VL-8BEvaluation Setting=Visual-Native Agent Harness2026.054.2
Qwen3-VL-30BEvaluation Setting=Direct Reasoning2026.053.8
Qwen3-VL-8BEvaluation Setting=Direct Reasoning2026.052.8
Claude-4-SonnetEvaluation Setting=Direct Reasoning2026.052