Mistral AI launches OCR 3 in next leg of expansion

Mistral AI introduced Mistral OCR 3, a model designed to extract text and embedded images from a wide range of documents, available today.

The company said in a statement that Mistral OCR 3 posts a 74% overall win rate over Mistral OCR 2 on forms, scanned documents, complex tables and handwriting. The company added that the model delivers state-of-the-art accuracy compared with both enterprise document-processing solutions and AI-native OCR offerings.

Mistral said the model supports markdown output enriched with HTML-based table reconstruction and emits HTML table tags with colspan and rowspan to preserve layout. Developers can access the model mistral-ocr-2512 via API, and the company noted pricing at $2 per 1,000 pages with a 50% Batch-API discount reducing the cost to $1 per 1,000 pages.

The firm said Mistral OCR 3 improves handwriting interpretation, forms detection, robustness to compression artifacts and low DPI, and complex table reconstruction. “OCR remains foundational for enabling generative AI and agentic AI,” Tim Law, IDC Director of Research for AI and Automation, said. “Those organizations that can efficiently and cost-effectively extract text and embedded images with high fidelity will unlock value and will gain a competitive advantage from their data by providing richer context.”

Access is available through the API and the Document AI Playground interface in Mistral AI Studio, the company said, and Mistral OCR 3 is fully backward compatible with Mistral OCR 2; documentation is at mistral.ai/docs.

The Recap

• Mistral AI introduced Mistral OCR 3 for enterprise and developer document processing.
• Pricing starts at $2 per 1,000 pages with discounts available.
• The model is available through API and Document AI Playground.