Image to Text
Processed on our servers. Nothing is stored after your request completes.
Save this tool
Keep Image to Text handy
Use your browser's Share menu, then “Add to Home Screen”.
Read the text out of a photo — receipts, screenshots, handwriting.
This tool runs on paid infrastructure and is available on Pro. See what Pro includes
- Drop in a photo or screenshot
- Choose whether you want plain text, Markdown, or a table as CSV
- Read it, then copy or download the result
Reading an image costs 2 AI credits, included with Pro.
About this tool
Text that only exists as a picture is text you cannot search, copy or edit. A screenshot of an error message, a receipt you need the figures from, a page of a book, a form filled in by hand — all of it has to be retyped, and retyping is where the mistakes come in. This reads the image and gives you the text back. It uses a vision model rather than a traditional character-recognition engine, and that choice is about the images people actually have. Classical engines want a flat, well-lit, high-contrast scan; what turns up is a phone photograph taken at an angle in a kitchen, on paper that is not flat, and that is precisely the case a model handles better. It also means nothing has to be downloaded before you start. There are three shapes of answer: plain text when you want the words, Markdown when the layout has headings and lists worth keeping, and CSV when you have photographed a table or a receipt and want the figures in a spreadsheet. This is the one tool here that costs AI credits, because every run is a real call to a model, and the price is shown before you press the button rather than after.
Frequently asked questions
Does this work on handwriting?
Often, and much better than a traditional engine would. Clear handwriting on a reasonable photograph usually comes out well; cramped or unusual hands are less reliable. It is worth checking the result against the original either way, which is true of any transcription.
Why does this one cost credits when everything else here is free?
Because every run is a genuine call to a model that costs real money, unlike the other tools, which run on your own device and cost us nothing. Credits come with a Pro account, and the price of a run is shown before you start it.
Is my image uploaded?
Yes, and this is the only tool here where that is true. Reading an image with a model means sending it to one, so the file goes to our server and on to the model provider. Every other tool on this site works entirely in your browser, and the difference is stated on each of them.
Can it read a table into a spreadsheet?
Yes, that is what the table mode is for. It returns CSV with the column headings in the first row, which you can open directly in a spreadsheet or run through the CSV to JSON tool. It works well on receipts, statements and printed tables.
Do I need to say what language the text is in?
No. The language field is a hint that helps with ambiguous or low-quality images, not a setting that restricts what can be read. Leaving it empty is fine for almost everything.
Help improve this tool
Found a bug, want a change, or need a tool we don't have?
You have an unsent draft.
Thanks — that's with us. You'll find the reply in your inbox.Go to my messages