What is 2markdown?
2markdown provides a developer-first API solution for converting web content and PDF documents into LLM-ready markdown. This service intelligently extracts relevant content, filters out noise, and preserves the original document structure, even with complex layouts.
The output is specifically formatted for optimal LLM consumption, reducing token usage and improving AI model understanding. The RESTful API offers native LangChain integration and OpenAI function support, facilitating rapid implementation.
Features
- Developer-First API: RESTful API with native LangChain integration and OpenAI function support.
- Intelligent Content Extraction: Cleans and extracts relevant content from websites and PDFs, filtering out noise and preserving structure.
- LLM-Optimized Output: Formats output specifically for LLM consumption, reducing token usage and improving understanding.
- Rapid Implementation: Allows developers to quickly integrate content processing into their AI applications.
Use Cases
- Building RAG applications
- Creating document-based chat interfaces
- Developing AI training pipelines
- Processing documents for LLMs
FAQs
-
What counts as a request/conversion?
1 URL = 1 request. For PDFs, we count the number of pages. Some endpoints have different costs, but it is all listed in the API docs. -
How much does PDF processing cost?
PDF processing is charged per page at $0.01 (1 cent) per page. For example, a 10-page PDF would cost $0.10 to convert. -
Do failed conversions count towards my quota?
Nope! We only count successful conversions. If something fails (bad URL, timeout, etc.), your quota stays untouched. -
How do you handle tables and complex layouts?
We preserve tables using standard Markdown syntax (the GitHub-flavored kind). For complex stuff like nested layouts, we use smart heuristics and OCR to maintain document structure. Lists, code blocks, quotes - they all get converted while keeping their hierarchy intact. -
Do you comply with GDPR/CCPA?
Yep, we're fully compliant. we don't store your data, we don't use it for training, and we don't do anything sneaky with it. It comes in, gets converted, goes back to you, and disappears from our system.
