Baidu Open-Sources Unlimited-OCR: A 3B-Parameter Model That Reads Entire PDFs in One Forward Pass
Sign in✓ Signed in
Baidu has released Unlimited-OCR, an open-source optical character recognition model that transcribes entire multi-page PDF documents in a single forward pass, eliminating the need for page-by-page processing. The model—a 3-billion-parameter mixture-of-experts system with 500 million active parameters—was open-sourced on June 22, 2026, under an MIT license, making it freely available for commercial and research use.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...