March 7 News.Mistral AI announced the launch of Mistral OCR on June 6, local time. APIIt also claims that the API has a good understanding of the OCR optics.character recognitionThe best performance in the field.
Mistral AI says the API has excellent multimodal document processing capabilities, with the ability to process documents from PDFs, slides, mathematical expressions, LaTeX academic documents, and more!Extracting text content from complex document elementsThe results are clearly presented in a Markdown file.
Mistral OCR ModelParses, understands and transcribes thousands of scripts, fonts and languages. According to data given by Mistral AI, the current version of the model outperforms a range of competitors such as Google Document AI, Microsoft Azure OCR, OpenAI GPT-4o and others in recognition.
Mistral AI has adopted Mistral OCR as the default document understanding model for its Le Chat chatbot, an API of thePriced at $1 per 1,000 pages(Note: Currently about 7.2 RMB), a single node can process 2000 pages per minute, and this model has been used in scenarios such as historical and cultural heritage preservation.
