On Thursday French large language model (LLM) developer Mistral launched a new API for developers who handle complex PDF documents. Mistral OCR is an optical character recognition (OCR) API that can turn any PDF into a text file to make it easier for AI models to ingest. LLMs that underpin popular GenAI tools, like OpenAI’s […] © 2024 TechCrunch. All rights reserved. For personal use only.
Mistral, a French developer of large language models (LLMs), has launched Mistral OCR, a new API designed to convert PDF documents into AI-ready Markdown files. Unlike traditional OCR tools, Mistral OCR is multimodal, detecting and formatting text alongside graphical elements like illustrations and photos. The API outputs text in Markdown, a format favored by AI models for its readability and structure. Mistral OCR is available through Mistral's API platform and major cloud providers, with on-premise deployment options for sensitive data. The company claims it outperforms OCR APIs from Google, Microsoft, and OpenAI, particularly with complex documents and non-English languages. It is designed to enhance AI workflows, such as Retrieval-Augmented Generation (RAG) systems, helping organizations process and analyze vast amounts of internal documentation efficiently.
