ResearchBenchmarksVisual Reasoning on Table SplitFollow77AccuracyGPT-4o5.86424.33242.861.268May 25, 2025Evaluation ResultsMethodMethodLinksAccuracyGPT-4oInference Mode=Tool UseInference Mode=Tool Use2025.0577GPT-4oInference Mode=Pure RunInference Mode=Pure Run2025.0575.7VTool-R1Model Size=7BModel Size=7B2025.0571.7VTool-R1Model Size=3BModel Size=3B2025.0557.9R1-VLModel Size=7BModel Size=7B2025.0545.4R1-VLModel Size=2BModel Size=2B2025.058.6