Every plan reads with the same Unlimited OCR 2 model. Plans differ in pages per month, pages per document and how many files you can read at once.
For occasional scans and short reports
Billed yearly · 33% off
For regular work with long PDFs
Billed yearly · 33% off
For archives, books and team volumes
Billed yearly · 33% off
Pay once, no subscription. Pack credits never expire and are used after plan credits.
A handful of documents
A project's worth of scans
Books, archives and backlogs
All prices in US dollars. Any tax is added at checkout.
FAQ
Files, pricing, privacy and the model. Still stuck? Write to us from the contact page.
Unlimited OCR 2 is an online OCR tool for long documents. You upload a scanned PDF or an image, and the open-source Unlimited OCR 2 model reads the pages in order and returns Markdown or plain text with headings, tables and formulas kept. It runs in your browser with nothing to install.
Unlimited OCR 2 keeps headings, paragraphs, lists, tables (as Markdown tables), formulas (as LaTeX) and the reading order of multi-column pages. Pictures are not exported as image files; their captions and any text inside them are kept as text.
Unlimited OCR 2 accepts PDF, PNG, JPG, WEBP. PDFs up to 50 MB, images up to 20 MB, one file per job. A document can have up to 50 pages with credit packs and up to 800 pages on the largest plan; for longer files, read them in page ranges. Password-protected PDFs are not supported.
Unlimited OCR 2 usually returns a single image in a few seconds. Long PDFs are read page by page while a counter shows progress; a few hundred pages can take several minutes. You can leave the page open or come back later: the finished result waits in your dashboard.
There is no free tier. Every page you read costs 1 credit and an image counts as one page, so a 40-page PDF costs 40 credits. Credits come from a monthly or yearly plan, or from one-time packs that never expire. The cost is shown before you start.
Credits are taken when a job starts. If Unlimited OCR 2 cannot read the file or the job fails, the full amount returns to your balance automatically and the job is marked as failed in your history. Scans that are very blurry or rotated may give weaker text, so check the preview before you rely on it.
Uploaded files and results are stored privately in your account and are only used to run your Unlimited OCR 2 job. They are never shared, published or used to train models. You can delete a job, together with its source file and result, from your dashboard at any time.
Yes. The text you extract with Unlimited OCR 2 belongs to you and can be used in commercial work. There is no watermark and no attribution requirement. You are responsible for having the rights to the documents you upload.
Unlimited OCR 2, a 3-billion-parameter open-source OCR model released by Baidu under the MIT license and designed for documents of hundreds of pages. This website is an independent service that runs the open model; it is not affiliated with Baidu.