pig4cloud/markitdown repository overview

⁠markitdown api server

microsoft markitdown api server

⁠Environment Variables

The service supports the following environment variables:

⁠Model Configuration

⁠OpenAI GPT-4V

• MODEL=gpt-4o
• BASE_URL=https://api.openai.com/v1⁠

⁠Docker Usage

⁠1. Quick Start

docker run -p 8000:8000 pig4cloud/markitdown

⁠2. Custom LLM Configuration

docker run -d \
  -p 8000:8000 \
  -e API_KEY=your_openai_api_key \
  -e MODEL=gpt-4o \
  -e BASE_URL=https://api.openai.com/v1 \
  pig4cloud/markitdown

⁠API Documentation

⁠Upload and Extract Text

Endpoint: POST /upload/

Request Format: multipart/form-data

Parameters:

• file: Image file

Response Example:

{
    "new_filename": "1679012345.jpg",
    "content": "extracted text content"
}

⁠Important Notes

1. Make sure you have a valid OpenAI API key with GPT-4V access
2. The service requires internet connectivity to access OpenAI's API
3. Uploaded files are automatically deleted after processing (default: 10 minutes)
4. The service listens on port 8000 by default
5. GPT-4V has a maximum input image size limit, ensure your images meet OpenAI's requirements

⁠System Requirements

• Docker
• Internet connection
• Valid OpenAI API key with GPT-4V access

⁠Features

• High-accuracy text extraction using GPT-4V
• Support for various image formats
• Automatic file cleanup
• Simple REST API interface
• Docker-based deployment

⁠Limitations

• Depends on OpenAI API availability
• Subject to OpenAI's rate limits and pricing
• Maximum image size limitations apply
• Internet connection required