Firecrawl Rewrites PDF Parser in Rust, Achieving Speed Improvements of Up to 5.7x

ME News reports that on April 15 (UTC+8), according to 1M AI News monitoring, the web data extraction tool Firecrawl has launched Fire-PDF—a PDF parsing engine rewritten in Rust—that accelerates the conversion of PDFs into structured Markdown by 3.5 to 5.7 times compared to its predecessor, with an average processing time of less than 400 milliseconds per page. The performance gain stems from minimizing unnecessary GPU calls. Firecrawl has also open-sourced the Rust library pdf-inspector, which classifies each PDF page in milliseconds: pure text pages are extracted natively without GPU usage, while only pages containing scanned documents or image-heavy content are processed through neural network layout models and the GLM-OCR visual language model. For example, in a 150-page financial report with 60 scanned pages, most pages require no GPU processing. In terms of accuracy, Fire-PDF applies tailored parameters for different content types: tables receive higher token limits and up to 25 seconds of generation time; formulas are preserved in LaTeX; and multi-column layouts use neural networks to predict reading order. Fire-PDF is now automatically enabled for all Firecrawl users with no configuration required. (Source: BlockBeats)