Mistral OCR 3 is an AI Productivity tool. AI model that extracts text and images from documents with structure preservation. Key features include Handwriting Recognition, Complex Table Reconstruction, and Structured Output Formats. Best for data scientists and analysts, legal professionals and software developers and engineers.
About Mistral OCR 3
Key Features
Handwriting Recognition.
Complex Table Reconstruction.
Structured Output Formats.
Low-Quality Document Processing.
Batch Processing API.
Document AI Playground.
Frequently Asked Questions
Mistral OCR 3 works with a wide range of documents including invoices, receipts, forms, contracts, scanned archives, technical reports, and handwritten notes. It handles both clean scans and low-quality documents with compression artifacts or distortion.
Mistral OCR 3 costs $2 per 1,000 pages for standard API usage. If you use the Batch API for high-volume processing, the price drops to $1 per 1,000 pages. There's also a free plan available with usage data collection.
Yes. Mistral OCR 3 outputs markdown with HTML-based table reconstruction by default. You can also define a JSON schema to extract structured data from forms and tables into specific fields, which is helpful for automated pipelines.
In internal benchmarks, Mistral OCR 3 achieved a 74% win rate over its predecessor on forms, scanned documents, complex tables, and handwriting. Third-party tests show it outperforms Azure AI and AWS Textract on handwriting and table extraction tasks.
