unsloth/Nanonets-OCR-s

VISIONPricing:Input $0.32 / Cached $0.016 / Output $1.6Concurrent Unit Cost:1Model Size:3BQuant:BF16Context Size:32kPublished:Jun 16, 2025Architecture:Transformer0.0K Featherless Exclusive Cold

Nanonets-OCR-s by unsloth is a 3 billion parameter image-to-markdown OCR model designed for advanced document processing. It excels at transforming documents into structured markdown with intelligent content recognition and semantic tagging, making it highly suitable for downstream Large Language Model (LLM) processing. Key features include LaTeX equation recognition, intelligent image description, signature detection, watermark extraction, smart checkbox handling, and complex table extraction. This model is optimized for accurately converting diverse document elements into structured formats.

Loading preview...

Nanonets-OCR-s: Advanced Document-to-Markdown Conversion

Nanonets-OCR-s is a 3 billion parameter image-to-markdown Optical Character Recognition (OCR) model developed by unsloth. It specializes in converting complex documents into structured markdown, making the output readily consumable by Large Language Models (LLMs).

Key Capabilities

  • LaTeX Equation Recognition: Automatically converts mathematical equations into correctly formatted LaTeX syntax, distinguishing between inline and display equations.
  • Intelligent Image Description: Generates structured <img> tags with descriptions for images, including logos, charts, and graphs, detailing their content and context.
  • Signature Detection & Isolation: Identifies and isolates signatures within documents, outputting them in a <signature> tag for legal and business document processing.
  • Watermark Extraction: Detects and extracts watermark text, placing it within a <watermark> tag.
  • Smart Checkbox Handling: Standardizes checkbox and radio button representations using Unicode symbols (☐, ☑, ☒).
  • Complex Table Extraction: Accurately extracts and converts intricate tables into both markdown and HTML formats.

Good For

  • Developers needing to preprocess diverse documents for LLM ingestion.
  • Applications requiring structured extraction of complex document elements like equations, tables, and images.
  • Automating the conversion of legal, scientific, or business documents into machine-readable, semantically tagged formats.

Finetunes
2 models