Powered by the MiniCPM-OCR 1.0 document model
Drop a scan, photo or PDF and MiniCPM-OCR 1.0 returns Markdown, plain text and JSON. Tables, formulas, headings and reading order stay intact.
No free tier · plans from $10.42/mo · 1 credit = 1 page
Your extracted text will appear here.
doc_title
MiniCPM-OCR 1.0 is a document parsing model that reads images and PDFs and returns structured text: Markdown for every page, plus layout blocks with labels, reading order and coordinates.
This site runs MiniCPM-OCR 1.0 for you in the browser. There is nothing to install and no GPU to rent: upload a contract, paper, invoice, exam sheet or newspaper page and MiniCPM-OCR 1.0 tells titles, paragraphs, tables and formulas apart. Paste the Markdown into Notion, Obsidian or Google Docs, or feed the JSON from MiniCPM-OCR 1.0 into your own pipeline.
MiniCPM-OCR 1.0 rebuilds rows, columns and merged cells as HTML tables inside the Markdown, so statements and spec sheets keep their grid.
Equations in papers, textbooks and exams come out of MiniCPM-OCR 1.0 as LaTeX you can paste into Overleaf or any Markdown editor.
Two-column papers and newspapers are read in order. Headers, footers and page numbers stay out of the Markdown but remain in the JSON.
MiniCPM-OCR 1.0 can straighten sideways pages before reading them and read the text inside stamps and seals on contracts and forms.
ordered_list
Three steps, no Python and no GPU.
Drop a PDF of up to 100 pages, or a PNG, JPG, WEBP, BMP or GIF image, into the MiniCPM-OCR 1.0 workbench above.
Turn auto-straighten and stamp reading on or off. The page count and credit cost show before you start.
Click Extract text. MiniCPM-OCR 1.0 returns Markdown, plain text, a layout view and JSON that you can copy or download.
pricing_table
Every plan includes the full MiniCPM-OCR 1.0 workbench. Pick the pages you need each month, or buy a credit pack.
For occasional scans and receipts
billed yearly · $0.017 / page
For weekly reports, papers and invoices
billed yearly · $0.017 / page
For teams and heavy document workflows
billed yearly · $0.017 / page
faq
Answers about results, pricing, files and privacy.
MiniCPM-OCR 1.0 is a document parsing model for images and PDFs. It returns page Markdown with HTML tables and LaTeX formulas, plus labeled layout blocks with coordinates. This site is an online workbench for MiniCPM-OCR 1.0: upload a file and download the result, no setup needed.
MiniCPM-OCR 1.0 extracts printed text, titles, lists, tables (as HTML tables in Markdown) and math formulas (as LaTeX), in reading order. It can also read stamps and seals. Pictures are skipped, so results are text only.
Sign in, drop your file into the MiniCPM-OCR 1.0 workbench at the top of this page and click Extract text. A single image usually takes a few seconds; a long PDF can take a few minutes. You can leave the page: the result is saved to your History.
There is no free plan or trial. Every page costs 1 credit, and an image counts as one page. Subscriptions add credits every month; unused subscription credits expire at the end of each monthly cycle. Credit packs are a one-time purchase and never expire.
MiniCPM-OCR 1.0 reads PDF files with up to 100 pages and PNG, JPG, WEBP, BMP and GIF images. Each file can be up to 20 MB. Password-protected PDFs are not supported.
Yes. You own your documents and the extracted text, and you can use it in commercial work on any plan. No watermark is added.
Files travel over HTTPS and are used only to process your request; this site does not keep your uploaded file. Extracted text is kept in your History for 15 days so you can download it again, then deleted automatically. Download anything you want to keep before then.
If MiniCPM-OCR 1.0 cannot process a file, the credits for it are refunded automatically. Try a sharper scan, or split very large PDFs into smaller parts.
Scans, photos and PDFs become Markdown, text and JSON in seconds.
Open the MiniCPM-OCR 1.0 workbench