Accurate OCR for notes, forms, tables and handwriting
- We’re committed to empower the AI community with open technology. Our open models sets the bar for efficiency, and are available for free, with fully permissive license. - Our optimized commercial models are designed for performance and are available via our flexible deployment options.
By extracting text and structured information from documents automatically, OCR 3 eliminates manual transcription work. Users can process large document volumes programmatically, freeing teams to focus on analysis and decision-making rather than data entry.
OCR 3 is available as a Premier API model (mistral-ocr-latest) that integrates into applications via Mistral's API infrastructure. Supports deployment options including self-hosted, Mistral cloud (EU-hosted), and major cloud providers (AWS, Azure, Google Cloud, Snowflake, etc.) for flexible enterprise integration.
OCR 3 provides both OCR ($2/1000 pages) and annotation services ($3/1000 pages), allowing users to build end-to-end document automation pipelines. Integration with Mistral Studio and APIs enables orchestration of multi-step document workflows with observability and audit trails.
Mistral OCR 3 is described as 'the world's best document extraction and understanding model' with multimodal capabilities. It enables users to convert PDFs, images, and scanned documents into structured, machine-readable data via API, supporting batch processing with 50% discount for cost optimization.