PDF Compression for AI/LLM Workflows
Reduce PDF token count for LLMs and RAG pipelines. Strip metadata, compress images, and remove PDF junk to lower embedding costs and improve AI workflow efficiency.
Why compress PDFs for AI?
- Lower token costs — Smaller files mean fewer tokens for embeddings and inference.
- Strip metadata and junk — Remove EXIF, annotations, and hidden data that adds noise to LLM inputs.
- Faster RAG pipelines — Compressed PDFs process faster in document ingestion and chunking workflows.
How it works
Call the PDF Squeezer API with stripMetadata=true and convertToJpeg=true before feeding PDFs to your embedding model or LLM. See the metadata removal guide for details.