nanonets/Nanonets-OCR-s

VISIONPricing:Input $0.32 / Cached $0.016 / Output $1.6Concurrent Unit Cost:1Model Size:3BQuant:BF16Context Size:32kPublished:Jun 10, 2025Architecture:Transformer1.6K Featherless Exclusive Cold

Nanonets-OCR-s by Nanonets is a 3 billion parameter image-to-markdown OCR model designed for structured document processing. It excels at converting complex documents into markdown with intelligent content recognition, including LaTeX equation recognition, image description, signature detection, watermark extraction, and smart checkbox handling. This model is primarily used for preparing diverse document types for downstream processing by Large Language Models (LLMs), offering a context length of 32768 tokens.

Loading preview...

Nanonets-OCR-s: Advanced Image-to-Markdown OCR

Nanonets-OCR-s, developed by Nanonets, is a 3 billion parameter image-to-markdown OCR model that specializes in transforming complex documents into structured markdown. Unlike traditional OCR, this model focuses on intelligent content recognition and semantic tagging, making the output highly suitable for further processing by Large Language Models (LLMs).

Key Capabilities

  • LaTeX Equation Recognition: Automatically converts mathematical equations into properly formatted LaTeX syntax, distinguishing between inline and display equations.
  • Intelligent Image Description: Generates structured <img> tags with descriptions for images within documents, detailing content, style, and context for LLM consumption.
  • Signature Detection & Isolation: Identifies and isolates signatures, outputting them within a <signature> tag, crucial for legal and business documents.
  • Watermark Extraction: Detects and extracts watermark text, placing it within a <watermark> tag.
  • Smart Checkbox Handling: Converts checkboxes and radio buttons into standardized Unicode symbols (☐, ☑, ☒) for consistent processing.
  • Complex Table Extraction: Accurately extracts and converts complex tables into both markdown and HTML formats.

Good for

  • Document Digitization: Converting scanned documents, PDFs, and images into structured, machine-readable formats.
  • LLM Pre-processing: Preparing diverse document types (e.g., scientific papers, legal contracts, financial reports) for ingestion and analysis by large language models.
  • Automated Data Extraction: Extracting specific entities like equations, signatures, watermarks, and table data from documents with high fidelity.
  • Form Processing: Standardizing the output of checkboxes and radio buttons from forms for reliable automated processing.

Finetunes
2 models