Mistral AI has unveiled what it claims is the world's most advanced document understanding technology. Dubbed "Mistral OCR," the new optical character recognition API aims to transform how organizations extract and utilize information from their vast document repositories.

With approximately 90% of the world's organizational data is locked in document format, Mistral OCR promises to unlock this potential by intelligently processing and understanding complex document elements, including interleaved imagery, mathematical expressions, tables, and advanced layouts such as LaTeX formatting. The model enables deeper understanding of rich documents such as scientific papers with charts, graphs, equations and figures.

"Unlike other models, Mistral OCR comprehends each element of documents with unprecedented accuracy and cognition," a company spokesperson stated. The technology processes images and PDFs as input and extracts content as ordered interleaved text and images.

What sets Mistral OCR apart from competitors is its multilingual capability, processing thousands of scripts, fonts, and languages across all continents. Internal benchmarks suggest the system outperforms offerings from major tech companies including Google Document AI, Azure OCR, and Gemini models across various metrics.

The company reports that in testing, Mistral OCR achieved a 94.89% overall accuracy rate, compared to Google Document AI's 83.42% and GPT-4o's 89.77%. The technology showed particular strength in processing mathematical content, scanned documents, and complex tables.

Speed appears to be another competitive advantage, with the company claiming it can process up to 2,000 pages per minute on a single node - reportedly faster than similar solutions in the market.

The API is available immediately on Mistral's developer platform at a rate of 1,000 pages per US dollar, with approximately double the pages per dollar available through batch processing. The company is also making the technology available for self-hosting on a selective basis for organizations dealing with sensitive or classified information.

Early adopters have reportedly used the technology for digitizing scientific research, preserving historical documents, streamlining customer service, and converting various forms of literature into AI-ready formats.

The API is available today on Mistral’s developer suite la Plateforme, and coming soon to cloud and inference partners, as well as on-premises.