Re-OCR of 2,724-page Encyclopaedia Britannica costs ~$3.40 with Baidu's Unlimited-OCR-2

vanstriendaniel · x · 2026-10-27

A developer re-OCR'd the complete 1771 Encyclopaedia Britannica (2,724 pages) with Baidu's new Unlimited-OCR-2. The model reads many pages per request (Baidu demos a whole book in one go), so sentences split across page breaks come back intact. Cost: $0.0012/page, $3.40 total via one command on Hugging Face Jobs — down from $5 in February using a 0.9B GLM-OCR on an L4 GPU. The screenshot shows a cross-page sentence correctly joined.

Related event: Baidu's Unlimited-OCR-2 Rescans 2,724-Page Encyclopaedia Britannica for $3.4(2 posts)→

Original post →

More from Multimodal

Multimodal channel →