Baidu Open-Sources Unlimited-OCR for Long Documents
thisdudelikesAI · x · 2026-07-20
Baidu’s Unlimited-OCR has been open-sourced as a local document OCR model.
The post says it is a 3B model that can process long PDFs in a single pass with a 32K context window, supports multiple languages, and runs entirely on local hardware. It is claimed to reach 93% on a standard parsing benchmark, keep error rates below 0.11 after 40 pages, and work with Transformers, vLLM, SGLang, Docker, Ollama, and llama.cpp.
The comparison point is that cloud OCR services charge per page, while this setup avoids upload limits, per-page billing, and sending private documents to a remote server. The post also says it has surpassed 1.9 million downloads on Hugging Face.
Related event: Baidu Open-Sources Unlimited-OCR for Long Documents(4 posts)→
More from Infra
- Cloudflare Workers adds spanContext, setStatus and updateName tracing APIs — irvinebroque · 2026-10-27
- Stranded compute, neocloud debt worries, and the rise of 'micro-neoclouds' — mobileraj · 2026-10-27
- Chutes roadmap: cut loss-making inference, push hourly rentals and edge nodes — jon_durbin · 2026-10-27
- HBM market to hit $106B by 2027, up 26X from $4B in 2023, per Bain — Beth_Kindig · 2026-10-27
- AI boom fuels a cloud computing 'land grab' as hyperscalers scramble to keep pace — DavidLinthicum · 2026-10-27
- Gemma hits 2.85x speedup on vLLM: 37,900 to 108,000+ tokens/sec on one A100 — patloeber · 2026-10-27