Mistral OCR

Mistral OCR is an API that converts PDFs and images into structured markdown, keeping tables, equations and layout. It is designed for feeding documents to language models without losing formatting.

Speed and price per page are its main selling points, and it can process thousands of pages quickly. It also returns embedded images, which helps with reports and slide decks.

Quality is strong on many document types, though very noisy scans can still trip it. A good first thing to test if you are building document Q&A.