Skip to content

Parse and review documents in the WebUI

Open the online demo, or install MinerU and start the local interface from a terminal.

mineru-kit webui

Open the address printed in the terminal. By default the WebUI starts a local Router and one parsing worker. To submit files to an existing V1 service, run mineru-kit webui --api-url http://127.0.0.1:8000.

Upload and select pages

  1. Choose or drop a file in the upload area, or use Paste for a file or image on your clipboard. You can start reading a local PDF as soon as you select it.
  2. For a PDF or image, choose one of the service's available tiers. Native office documents, web pages, and other native formats use Flash and are parsed in full.
  3. To parse consecutive PDF pages, drag the two ends of the page range. Page numbers start at 1 and include both endpoints. All pages are selected initially, up to any --max-pages limit set by the deployment. Turning a preview page only changes what you see.
  4. Click Convert. MinerU automatically decides whether a PDF needs OCR. To ignore the text layer and recognize the pages again, turn on Force OCR before conversion.

Check the result against the source

Open the Markdown tab to review the formatted result. Use Contents to jump to a heading. For PDFs and images, click a region in the source to highlight its parsed content. Click a paragraph, heading, formula, or figure with a source location to return to that region. For a paragraph or table merged across pages, use Source page to switch between its source pages.

Region boxes are drawn over the source preview. Even after parsing a selection, you can still browse the full PDF using its original page numbers.

Download the result

Open Download in the results pane and choose a format. Markdown and JSON each download as a separate ZIP containing the content and its images/ assets. The JSON tab and its download contain Structured Content. For Middle JSON, use the Tasks default bundle or mineru-kit parse --format middle_json.

The menu also offers HTML, DOCX, LaTeX, EPUB, and PDF. LaTeX includes the .tex file and images. The PDF is rendered from the parsed result for sharing. Standalone HTML references images served by the running WebUI. For an offline copy with its images, choose the Markdown ZIP. Output formats by interface