markdrop
Convert PDFs to Markdown and HTML with AI image and table descriptions.
Extract tables and images from documents using six major LLM providers.
Markdrop is a Python package that converts PDFs into structured Markdown and interactive HTML while extracting images and tables. It is built for developers and professionals who need automated, AI-powered descriptions for visual elements within documents.
What it is
The package uses Docling for PDF conversion and Microsoft's Table Transformer for table detection. It then connects to Gemini, OpenAI, Anthropic Claude, Groq, OpenRouter, or LiteLLM to generate detailed text descriptions of extracted images and tables.
Who it's for
It suits developers and technical professionals handling heavy document processing pipelines requiring structured text output. It does not suit non-technical users who require a graphical user interface rather than a command line tool.
Key features
Multi-Provider AI Descriptions
Generate descriptions for images and tables using six major LLM providers including Gemini and Claude.
Interactive HTML Export
Convert documents into HTML containing downloadable Excel spreadsheets for extracted tables.
Fast CPU Mode
Run lightweight Markdown conversion using CPU-optimized modes for lower resource overhead.
Batch Image Processing
Analyze directory batches of images with multiple LLM clients using custom prompts.
Up and running in 3 steps
Our honest verdict
What's great
- Supports six major LLM providers with customizable vision and text models
- Extracts images via XRef IDs alongside Table Transformer detection
- Outputs interactive HTML with downloadable Excel tables
- Includes a fast CPU mode for lightweight document processing
Where it falls short
- Requires Python environment setup and command-line execution
- Relies on external paid API keys for the AI description features
- Local HuggingFace model usage requires installing extra dependencies
Pricing & license
Markdrop is free under its open source licence, but running the AI-powered description features requires API keys for the supported LLM providers.
Open remote AI-training roles
874 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
Use it, or have us set it up
FAQ
Which AI providers are supported?
Google Gemini, OpenAI, Anthropic Claude, Groq, OpenRouter, and LiteLLM are supported out of the box.
Can I use Markdrop without an AI API key?
Yes, you can run PDF to Markdown conversion locally, but AI image and table descriptions require a valid API key.
