Mistral OCR 4
Key point
Mistral AI has released Mistral OCR 4, which provides structured data and bounding boxes in addition to text extraction.
Details
Mistral AI has released Mistral OCR 4, which goes beyond simple text extraction to understand document structure. This model returns Bounding Box, block classification (headings, tables, formulas, signatures, etc.), and inline confidence scores alongside the text, providing structured data optimized for RAG and enterprise search pipelines.
Key features and performance:
- High-performance benchmarks: Recorded a score of 85.20 on OlmOCRBench, demonstrating higher performance than major OCR and Document AI systems.
- Extensive language support: Supports 10 language groups, totaling 170 languages, while maintaining high accuracy even for low-resource languages.
- Flexible deployment: A compact model that can run in a single container, supporting Self-hosted deployment for companies where data sovereignty is critical.
- RAG optimization: The extracted structured blocks refine the retrieval units of RAG systems through Semantic Chunking.
Pricing policy:
- API usage: $4 per 1,000 pages ($2 with a 50% discount when using the Batch API)
- Document AI (Mistral Studio): $5 per 1,000 pages
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.