Fast Image to Text (OCR)
Extract text from an image with the smallest OCR model — the quickest to download and run.
Drop images with text here or click to choose
Photos, scans and screenshots: JPG, PNG, WebP, GIF, BMP, TIFF or AVIF. Reads English and other Latin-alphabet languages, and Chinese. It can't read Korean or Japanese kana. Files are processed in your browser and never uploaded. They stay in this browser until you remove them.
Ctrl+OchooseCtrl+Vpaste
Text recognition model
Fastest model · …
Reads English and other Latin-alphabet languages (French, German, Spanish, Polish…) and Chinese. It can't read Korean, Japanese kana, Russian or other Cyrillic, Arabic, Thai, Hindi or Vietnamese.
Licenses: models (Apache-2.0) · engine (MIT)
Turn a screenshot, photo or scanned page into text you can copy and edit — quickly. This is the lightest of the three Image to Text tools: its recognition model is only 6 MB, so the first start is quick even on a phone, and after that a screenshot is read in a second or two. Everything runs in your browser: the image never leaves your device, and it's free with no sign-up.
Compare the three Image to Text models
| Image to Text (Fastest)This page | Image to Text (Balanced) | Image to Text (Accuracy) | |
|---|---|---|---|
| Model download | 6 MB | 29.7 MB | 132.3 MB |
| First use: download with the engine at 50 Mbit/s | 19.6 MB · ≈ 3 sec | 43.4 MB · ≈ 7 sec | 145.9 MB · ≈ 24 sec |
| After the first use | Kept in your browser — no second download | ||
| Speed | Fastest | Fast | Slowest |
| Accuracy | Good on clear text | Better on photos and small print | Best on hard images |
| Reads | Latin alphabet (English, French, German…) and Chinese | Latin alphabet (English, French, German…), Chinese, and Japanese | Latin alphabet (English, French, German…), Chinese, and Japanese |
The recognition engine (13.7 MB, or 27.4 MB in browsers with WebGPU) is downloaded once and shared by all three. Nothing is downloaded until you ask the tool to read an image.
How to convert an image to text quickly
- Drop one or more images anywhere on the page, paste a screenshot with Ctrl+V, or click to choose files. Images you added in another image tool are already in the list, and the sample receipt shows how it works.
- The first time, click Download model & read text. About 20 MB — the model and the engine that runs it — is downloaded once and saved in your browser; from then on, an image you add is read at once, the selected one as soon as you open the page, and the others already in the list when you click Read or Read all.
- Check the text beside the picture. Click a region in the image to find its line; lines the model is unsure about are underlined. Fix anything by typing, then click Copy text (Ctrl+Shift+C) or Download .txt.
- Optional: under Hide the text in the image, pick a color and download a copy of the picture with the text covered by boxes.
Features
- The smallest model: 6 MB, about 20 MB with its engine, downloaded once and then saved in your browser
- A screenshot read in a second or two once the model is saved
- Reads English and other Latin-alphabet languages, digits and Chinese
- Text in reading order, with the regions outlined and numbered on the image
- Edit, copy or download the text as .txt; lines the model is unsure about are underlined
- Hide the text in the image and download the masked copy
- Several images at once
Is it private?
Yes. Recognition runs entirely in your browser; the image and the text never leave your device, and the images stay in this browser until you remove them from the list. The only download is the model and its engine, from this site, the first time you use it.
Frequently asked questions
Why is the first run slower?
Because the first time, your browser downloads the recognition model (6 MB) and the engine that runs it (14 MB, or 27 MB in browsers that use the graphics card through WebGPU), then prepares them. Both are saved in the browser, so the next images and your next visit start with no download. The engine is shared with the Balanced and Accuracy tools, so trying those later downloads only their model.
Which languages can it read?
Printed English and other languages written in the Latin alphabet (French, German, Spanish, Portuguese, Italian, Polish, Czech, Turkish…), digits and symbols, and Chinese. It can't read Korean, Japanese kana, Russian or other Cyrillic, Arabic, Thai, Hindi or Vietnamese; for Japanese, use the Balanced or Accuracy tool.
How accurate is the fastest model?
Very good on clear printed text such as screenshots, documents and signs. For small print or blurry photos, the Balanced and Accuracy tools read more correctly, and lines the model is unsure about are underlined so you can check them.
Fastest, Balanced or Accuracy — which should I use?
Fastest, when the text is clear and you want the smallest download: about 20 MB the first time. Balanced (about 43 MB) reads photos and small text more accurately and also reads Japanese; Accuracy (about 146 MB) reads the hardest images best but is slow without graphics acceleration (WebGPU).
How do I free the space the model uses?
Click Delete downloaded models under Text recognition model: every saved model is removed from this browser. It is downloaded again the next time you read text.