LiteParse v2 Released: 100× Faster Open-Source PDF Parser in Rust

LlamaIndex released LiteParse v2, a Rust rewrite achieving up to 100× faster PDF parsing with sub-second processing for 450-page documents. It supports 50+ document types, runs in the browser via WASM, offers Python/TypeScript/Rust/WASM bindings, and surpassed 5,000 GitHub stars.

LiteParse v2 Released: 100× Faster Open-Source PDF Parser in Rust

LlamaIndex released LiteParse v2, a complete Rust rewrite of its document parsing library that achieves up to 100× speed improvement and sub-second processing for documents up to 450 pages and 100MB. The library supports more than 50 document types, compiles to WASM for browser and edge runtime usage, and provides native bindings for Python, TypeScript, Rust, and WASM. The repository crossed 5,000 GitHub stars on launch, and an integration for scientific agent workflows (local PDF parsing with grounded citations and OCR for scanned papers) was published simultaneously by K-Dense AI.

Why It Matters

Fast, local, multi-format document parsing is a baseline requirement for any production RAG or document-intelligence pipeline; LiteParse v2's WASM support makes it the first serious browser-native option, extending document AI to edge and offline use cases.