Unstructured's New Transform Models Beat GPT-6, Claude and Gemini at Parsing Tables, Forms and Scans
SACRAMENTO, Calif.--(BUSINESS WIRE)--Unstructured, whose open source software recently surpassed 100 million downloads, today launched new models for Transform, its document parsing solution for turning PDFs, scans and other complex documents into data AI agents can act on. In testing on 224 real-world business pages, GPT-6, Claude and Gemini produced up to 2.3 times as many table errors as Transform and invented about 50% more content than Unstructured that was never in the document.¹ The new m
Read the full story at Business WireOpen in the live monitor