Baidu's Unlimited OCR Handles Multi-Page Documents in One Read

Baidu released Unlimited OCR on 23 June 2026, a model that processes multi-page documents in a single pass by maintaining semantic context across pages. Conventional OCR systems fragment documents into chunks, losing information between pages; this architecture preserves that continuity, so details from page three can clarify content on page thirty. Code and weights are available on GitHub and Hugging Face.