Unstructured's New Transform Models Beat GPT-6, Claude and Gemini at Parsing Tables, Forms and Scans

The Quick Take
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 Transf
Ground Level News gathers headlines from public news feeds. This page shows a short summary only; the full article and all rights belong to Business Wire.



