DeepSeek OCR 2: Visual Causal Flow for Documents
DeepSeek OCR 2 is an open-source OCR and document understanding model built for complex layouts, Markdown output, and human-like reading order.
DeepSeek OCR: Open-Source OCR Model for Documents
DeepSeek OCR is an open-source vision-language OCR model that converts document images into structured text and Markdown with efficient visual token compression.