Image to Text

Image to Text / Markdown

Image to Markdown. Keep the useful parts.

Turn a JPG, PNG, or WebP into an editable .md file. Start with free browser OCR below, or switch to AI OCR for a document that needs structure. You choose the mode before any cloud conversion.

PRIVATE BY DEFAULT
01YOUR IMAGE

Drop your images here

or choose a file from your device

JPG, PNG, WebP · 10 MB each · up to 25 megapixels

No image handy? Try one:
02EXTRACTED TEXT Ready when you are
T

Your next copy & paste starts here.

Extract text, check it against the image, then copy or download.

.txt.md.json

Runs in your browser. Your images stay on your device.

No signup. No watermark. Yours to keep.Convert images in bulk →

Choose text extraction or structure recovery.

Free browser OCR

Free OCR reads printed text on your device. Exporting that text as Markdown gives you a .md file, but does not automatically rebuild headings, code fences, or table cells. This is useful for a quick note you plan to edit yourself.

Optional AI OCR

AI OCR sends the selected image to Replicate using the Marker model. It can infer document structure and return Markdown, but headings and reading order still need review. AI uses one credit per image; signup includes two trial images for the website.

How to convert an image to Markdown

  1. 01Add a sharp image. Crop to the document, keep the page upright, and select the correct language for free OCR. JPG, PNG, and WebP are supported, up to 10 MB and 25 megapixels each.
  2. 02Run OCR, then compare the output with the image. Check table columns, amounts, punctuation, and code indentation. Use AI for a second attempt when local text extraction loses too much layout.
  3. 03The Markdown tab is selected on this page. Edit the result, copy it, or export the .md file to your notes, repository, or documentation editor. API users retrieve result.markdown after the job finishes.

A readable screenshot is not always a faithful document.

A table may become plain lines; a code screenshot may lose spaces; handwriting can be ambiguous. We publish the original images and uncorrected results so you can judge the tradeoffs. The examples are small synthetic fixtures, not an accuracy guarantee for your documents.

See the actual OCR results →
Automate with the API, CLI, or MCP →