title: "Presentazione di Mistral OCR 3 | Mistral AI"

Mistral AI has released Mistral OCR 3, a new model for extracting text and images from various documents with high accuracy. It demonstrates a 74% overall win rate against its predecessor on forms, scanned documents, complex tables, and handwriting. The model supports enriched markdown output with HTML table reconstruction to preserve document structure. Developers can integrate it via API, and a Document AI Playground offers a user-friendly interface for analysis, with self-hosting options available for data privacy.