GitHub · by Jerry Liu

DocJev

Classify and split document packets with Jev. Open source and free by default.

Jerry Liu

@jerryjliu0

DocJev is the fastest way to classify and split complex document packets ⚡️. (and the default is fully free and OSS!) I made a sick teaser video below. TY Opus 5.5 🙏 check it out: github.com/jerryjliu/docj…

Sep 22, 2026 · 822 likesOpen on X

Jerry Liu’s library for document classification and splitting. You give it a document and some category rules in plain language, and Jev returns the document’s category, or the boundaries between the sub-documents inside a packet. He put it at 6x faster than gpt-5.6-luna at equivalent accuracy.

The OCR backend is a choice. liteparse is free and open source and is what the speedup above is measured with, preprocessing included, which makes it the option for documents that are already digital. LlamaParse is the VLM route for complex pages, and it adds latency at preprocessing time, so it pays off where the digitalised representation is cached and reused downstream.