MINICPM-OCR 1.0

From document pixels to useful content.

Recognize the content and structure of your pages, then check the result alongside the original before it enters your workflow.

Text & page layout

Read English and Chinese document pages, with headings, paragraphs and region coordinates that connect each result to its source.

Tables you can use

Review recognized table cells and merged structure. Export structured results for spreadsheets, reports and downstream workflows.

Formulas in LaTeX

Convert mathematical notation into editable LaTeX. Preview formulas alongside the original page to check the result.

Check every region against the source

Select a recognition box or a result block to locate its counterpart. Linked panels, reading-order labels and zoom controls help you check headings, table cells and equations in context.

Make corrections before you export

Double-click a recognized region to edit its content. Your reviewed text is used in Markdown, structured JSON and Word exports. OCR is not infallible; check important figures and formulas against the original.

Choose a format for the next step

Use Word for an editable document, Markdown for text and LaTeX workflows, or JSON for page dimensions, region coordinates, labels and reading order. These are result exports; plans do not include a public API.