How it works
Upload. Review. Export.
One workspace takes you from an image or PDF to reviewed, editable content. Start with one free document of up to three pages, without an account.
- 01
Choose your file and check the page count
Use the upload button, drag a file onto the workspace or paste an image. Supported images are PNG, JPEG, GIF, WebP and BMP, up to 24 MiB. Unlocked PDFs can be up to 64 MiB and 500 pages. The free experience covers one document of up to three pages; paid recognition uses one credit per page.
- 02
Start recognition and keep the workspace open
Choosing a file creates a local preview. Select Recognize document when you are ready to send it for processing. PDF pages are rendered in your browser and processed in sequence. Watch the page progress; you can cancel a running task. Failed or cancelled documents restore their allowance or credits.
- 03
Compare the original and correct the result
Select a source box or result region to locate the matching content in the other panel. Use Display to control recognition boxes, reading order and linked panels. Switch between Preview and Markdown, use the page controls for longer documents and double-click a region to correct its text.
- 04
Download the reviewed document
When recognition finishes, Export becomes available in the workspace action bar. Choose Word (.docx), Markdown (.md) or Structured JSON (.json). Exports include every page and your edits. Copy all copies the document as Markdown. These controls stay above the result while you review it.
Understand where your document goes
Your file stays in the browser until you request recognition. ModelBest processes recognition inputs through our server. Originals use private R2 storage with a 24-hour expiration rule; physical deletion follows the storage lifecycle schedule. Results and local recovery are available for up to 24 hours.
What happens to my document?