DeepSeek-OCR
DeepSeek-OCR is an open source model for Contexts Optical Compression, built to explore the boundaries of visual-text compression and investigate the role of vision encoders from an LLM-centric viewpoint. It is designed to compress long contexts through optical 2D mapping, using DeepEncoder as the core engine and DeepSeek3B-MoE-A570M as the decoder. DeepEncoder maintains low activations under high-resolution input while achieving high compression ratios, keeping the number of vision tokens manageable for document understanding. The model supports OCR and document parsing workflows for images and PDFs, with inference through vLLM or Transformers. Users can run image OCR with streaming output, process PDFs with high concurrency, or run batch evaluation for benchmarks. DeepSeek-OCR can convert documents to Markdown, perform free OCR without layouts, parse figures, describe images in detail, and locate referenced text inside an image.
Learn more
Unsiloed
Unsiloed AI is a document processing platform that turns PDFs, images, spreadsheets, scans, and other unstructured files into JSON and Markdown that LLMs and AI agents can use. The platform acts as a document layer for enterprise AI, helping teams parse, extract, and split complex documents without relying on brittle OCR pipelines. Its proprietary dual-stream vision models read both content and layout, preserving tables, figures, forms, signatures, handwriting, hierarchy, and document structure. Unsiloed can extract structured fields into JSON, convert documents into LLM-ready Markdown, and split multi-document files or long documents into retrievable chunks. The platform supports workflows across financial reports, legal contracts, invoices, healthcare records, regulatory filings, scanned documents, spreadsheets, and mixed-layout enterprise files.
Learn more
PaddleOCR
PaddleOCR is a leading open source OCR toolkit and document AI engine that turns PDFs and images into structured, LLM-ready data with high accuracy. It is designed to bridge the gap between documents and large language models by extracting, recognizing, parsing, and organizing information from scanned pages, photos, forms, tables, formulas, charts, and complex layouts. PaddleOCR supports more than 100 languages and provides a practical toolkit for building intelligent RAG and agentic applications that need reliable document understanding. Its core capabilities include PaddleOCR-VL, PP-OCRv5, PP-StructureV3, and PP-ChatOCRv4. PaddleOCR-VL is an ultra-compact vision-language model for multilingual document parsing, supporting 109 languages and performing well on complex elements such as text, tables, formulas, and charts. PP-OCRv5 is built for universal-scene text recognition.
Learn more
Mistral OCR 4
Mistral OCR 4 is a document extraction and understanding model built for enterprise search, RAG, domain-specific retrieval pipelines, and production-grade document intelligence. It extracts and structures content from a wide range of documents, moving beyond clean text and tables to return a structured representation of each page. Alongside extracted text, OCR 4 provides bounding boxes, typed-block classification, and inline confidence scores, helping downstream systems understand not only what the document says, but where each element sits, what role it plays, and how confident the model is in each region. Bounding boxes make in-context highlighting and reliable data pipelines possible, while block types and confidence scores support source-grounded citations, redactions, and human-in-the-loop verification. OCR 4 accepts common enterprise formats, including PDF, DOC, PPT, and OpenDocument, and supports 170 languages across 10 language groups.
Learn more
