Mistral OCR
mistral.ai
- Free plan
- Web
- API
- Self-hosted
What Mistral OCR is
Mistral OCR is a document-understanding and text-extraction service centered on OCR 4.1. It returns paragraph-level bounding boxes, structural block labels and confidence scores for individual blocks. OCR 4 handles 170 languages across 10 language groups and supports table extraction, structure extraction and handwriting OCR. Teams can call it through an API, submit batches through the Batch API, or use Document AI in Mistral Studio as a no-code interface. OCR 4 also serves as an ingestion component for Mistral Search Toolkit, an open-source framework for RAG and enterprise search. A single container can run on an organization’s infrastructure, with self-managed deployment available to enterprise customers; Mistral says this keeps document data in the customer environment. The OCR 4.1 API is listed at $4 / 1000 pages, and Document AI at $5 /1000 Annotated Pages. Python and TypeScript SDKs are available.
Who it suits
Teams extracting structured text from documents through APIs, batch jobs or a no-code studio. Enterprise organizations with residency or sovereignty requirements may consider self-hosting.
Good to know, and what to watch
- 170-language OCR coverage
- Bounding boxes, labels and confidence scores
- Batch API processing
- Table, structure and handwriting extraction
- Self-hosted container option
- Not intended for real-time processing
- Not for medical, legal or safety-critical decisions
- OCR 4.1 API costs $4 / 1000 pages
- Document AI costs $5 /1000 Annotated Pages
Plans
Every plan and priceOCR 4.1 API
Document AI
Free plan
The verdict
Mistral OCR offers structured extraction, broad language support and several deployment paths. Its stated exclusions make it unsuitable for high-stakes decisions and real-time workloads.
What the maker publishes
- Batch processing
- The OCR API supports batching through the Batch API. source
- Founded
- 2023 source
- Headquarters
- Paris, France source
- Integration
- OCR 4 is an ingestion component of Mistral Search Toolkit, an open-source search framework for RAG and enterprise search. source
- Languages
- OCR 4 supports 170 languages across 10 language groups. source
- Limitations
- Mistral says OCR 4 is not intended for medical diagnosis, legal advice or judgment, high-stakes financial decisions, safety-critical systems, real-time processing, or non-document inputs. source
- Privacy
- Mistral says self-hosting keeps document data in the customer’s environment for residency, sovereignty, and compliance needs. source
- Purpose
- OCR 4.1 is Mistral’s OCR service for document understanding and text extraction. source
- Self-hosting
- OCR 4 can run in a single container on an organization’s own infrastructure; self-managed deployment is available to enterprise customers. source
- Structured extraction
- OCR 4.1 provides paragraph-level bounding boxes, structural block labels, and block-level confidence scores. source
- Support
- Mistral’s Enterprise APIs include premium support and system-level SLAs. source
- Ways to use it
- OCR is available through an API, and teams can use Document AI in Mistral Studio as a no-code interface. source
Checked against the maker's pages on 29 Sept 2026.