MISTRAL RELEASES OCR 4 WITH ADVANCED DOCUMENT EXTRACTION

The new model advances structured document extraction by returning text alongside spatial metadata. Bounding boxes pinpoint extracted content locations, while block classification categorizes document sections. Inline confidence scores indicate extraction reliability for each element. The 170-language support positions OCR 4 as a multilingual alternative to existing document processing solutions. The combination of precise positioning and confidence metrics enables more sophisticated automation workflows in document digitization, data entry, and content management. Mistral AI, known for developing open and commercial large language models, continues expanding its product suite beyond traditional LLM applications. OCR capabilities integrate with document understanding tasks increasingly demanded by enterprises processing high-volume paperwork and digital scans.