A streamlined document processing library that leverages AI vision models for OCR and data extraction. Designed to handle complex document layouts including tables, charts, and irregular formatting.

  • Universal Document Support: Process PDFs, DOCX, images, and more
  • Vision-Based Processing: Uses AI vision models for superior layout understanding
  • Structured Output: Converts documents to clean Markdown or extracts specific data
  • Multi-Platform: Available for both Node.js and Python

​
Getting Started

Zerox is available as both a Node.js and Python package.

Node.js

Python

​
How It Works

  1. Upload any supported document
  2. Document is automatically converted into a series of images
  3. Vision models process each page
  4. Returns clean, structured output in Markdown or extracted data

Try out the hosted version here: https://getomni.ai/ocr-demo.

​
SDK Feature Support

Here’s the SDK feature support table in markdown format:

FeatureNode.jsPython
PDF Processing✓ (requires graphicsmagick)✓ (requires poppler)
Image Processing✓✓
OpenAI Support✓✓
Azure OpenAI Support✓✓
AWS Bedrock Support✓✓
Google Gemini Support✓✓
Vertex AI Support✗✓
Data Extraction✓ ( schema )✗
Per-page Extraction✓ ( extractPerPage )✗
Custom System Prompts✗✓ ( custom_system_prompt )
Maintain Format Option✓ ( maintainFormat )✓ ( maintain_format )
Async API✓✓
Error Handling Modes✓ ( errorMode )✗
Concurrent Processing✓ ( concurrency )✓ ( concurrency )
Temp Directory Management✓ ( tempDir )✓ ( temp_dir )
Page Selection✓ ( pagesToConvertAsImages )✓ ( select_pages )
Orientation Correction✓ ( correctOrientation )✗
Edge Trimming✓ ( trimEdges )✗
Node.js