Intended Use

Primary Use Cases

• Document-to-data, at scale. Convert physical documents (contracts, invoices, forms, and reports) to custom-structured digital copies in minutes.
• Extract and analyze. Enable AI-powered insights: detect patterns, validate data, and enhance enterprise search out of scanned documents.
• Translate and localize. Quickly localize contracts, reports, and correspondences across, with compliance-ready accuracy.
• Automate workflows with AI. Build end-to-end document pipelines — from OCR digitization to natural language querying, with fully automated structuring in-between.
• Monitor compliance and manage risk. Automatically audit document flows, redact sensitive data, or enforce retention policies, while keeping full traceability.

Basic OCR

Mistral Document AI comes with an improved Document OCR (Optical Character Recognition) processor, powered by our latest OCR model, mistral-ocr-2512, which enables you to extract text and structured content from PDFs and a variety of document types. Our latest model improves upon the previous OCR version (mistral-ocr-2507) with a 74% win-rate improvement on forms, scanned documents, complex tables, and handwriting.

Key Features

• Extracts text content while maintaining document structure and hierarchy
• Preserves formatting like headers, paragraphs, lists and tables
• Returns results in markdown format for easy parsing and rendering
• Handles complex layouts including multi-column text and mixed content
• Processes documents at scale with high accuracy
• Supports multiple document formats including:
  • image_url: png, jpeg/jpg, avif, png, tiff, gif, heic/heif, bmp, webp
  • document_url: pdf, pptx, docx, txt, epub, xml, rtf, odt, bib, fb2, ipynb, xml, tex, opml, man
    The OCR processor returns the extracted text content, images bboxes and metadata about the document structure, making it easy to work with the recognized content programmatically.

Go to Mistral AI OCR Documentation for further usage description.

Annotations

In addition to the basic OCR functionality, Mistral Document AI API adds the annotations functionality, which allows you to extract information in a structured json-format that you provide. Specifically, it offers two types of annotations:

• bbox_annotation: gives you the annotation of the bboxes extracted by the OCR model (charts/ figures etc) based on user requirement and provided bbox/image annotation format. The user may ask to describe/caption the figure for instance.
• document_annotation: returns the annotation of the entire document based on the provided document annotation format.

Key Features

• Labeling and annotating data
• Extraction and structuring of specific information from documents into a predefined JSON format
• Automation of data extraction to reduce manual entry and errors
• Efficient handling of large document volumes for enterprise-level applications

Go to Mistral AI Annotations Documentation for further usage description.

Limitations & Known Issues

• Mistral Document AI on Foundry can process documents up to 30Mb and 30 pages.
• Document Annotations are limited to 8 pages.
• While the pure OCR process performs efficiently and quickly, the annotation process can be slower from time to time and may result in timeouts. An optimization will be available in a few weeks timeline.

Supported languages:

de, fr, es, nl, it, pt, hu, pl, cs, da, ro, no, sv, id, th, vi, tl, ar, he, hi, bn, gu, kn, ta, te, ml, en, ru, uk, ko, ja, zh, tr, hy, ka

Preview Terms

This Azure Direct Model is a Preview and is subject to the Supplemental Terms of Use for Microsoft Azure Previews

Quality and performance evaluations

Top-tier benchmarks

To raise the bar, we introduced more challenging internal benchmarks based on real business use-case examples from customers. We then evaluated several models across the domains highlighted below, comparing their outputs to ground truth using fuzzy-match metric for accuracy.:

Model	Forms	Handwritten	Invoices	Complex Tables	Historical Scanned
DeepSeek OCR	82.6	57.2	70.5	84.4	81.1
Google Document AI	79.6	73.9	72.4	75.9	87.1
Azure OCR	86.2	78.2	80.2	85.9	83.7
AWS Textract	84.5	72.4	78.4	84.8	81.0
Mistral Document AI	95.9	88.9	91.8	96.6	96.7

Benchmarks per language

Language Group	DeepSeek OCR	Azure OCR	Google Doc AI	AWS Textract	Mistral Document AI
Chinese	90.5	87.0	83.3	N/A	97.1
East-Asian	90.7	89.9	80.7	N/A	97.6
Eastern Europe	90.3	94.5	91.4	88.9	98.6
English	94.6	93.5	91.6	93.9	98.6
Western Europe	94.6	94.3	92.2	94.3	98.8