Baidu Open-Sources Unlimited-OCR for Long Documents

Baidu has open-sourced Unlimited-OCR, a 3-billion-parameter model designed to process entire long documents at once, supporting up to 100 pages with a 32K context window. It activates only 500 million parameters during inference, enabling efficient local execution without page segmentation.

2026-07-20 ~ 2026-07-20 · 4 related posts

• Baidu Open-Sources Long-Doc OCR Model — brianrkelly · 2026-07-20
• Unlimited-OCR reads 100-page PDFs in one shot — Shruti_0810 · 2026-07-20

2 near-duplicate retellings: TheTimHayden · thisdudelikesAI