Private AI, on your device
Free AI tools that run in your browser
Each tool downloads its model only when you use it, and your browser may cache it. Your files, recordings and text stay on your device.
AI tools
AI Text Summarizer
Summarize text locally in your browser with an on-device AI model.
AI Writing Assistant
Paraphrase, clean up grammar, or shorten text with browser-based AI.
Image to Text (OCR)
Extract text from photos, screenshots and scanned PDFs on your device.
Speech to Text (Transcription)
Transcribe audio and video files or microphone recordings privately with Whisper.
Background Remover
Remove the background from portrait photos in your browser.
Blur Faces in Photos
Automatically find and blur or pixelate faces before you share a photo.
AI Alt Text Generator
Write alt text and image descriptions with an on-device vision model.
AI Translator
Translate between 13 languages, including English, Spanish, French, German, Chinese and Arabic, on your device.
Text to Speech
Read text aloud with voices installed on your device β nothing is sent online.
How on-device AI works
When you use one of these tools, your browser downloads an open AI model once and keeps it in its cache. A runtime compiled to WebAssembly, and on supported browsers WebGPU on your graphics card, then performs the calculations on your device. Your files and text are passed to the model inside the page; the model download contains no information about what you process.
Browser-sized models are smaller than large cloud services, so each one is built for a focused task. Speed depends on your computer or phone, and the output of any AI should be checked before you rely on it.
Models, download sizes and licences
| Tool | Model | First download | Licence |
|---|---|---|---|
| Image to Text (OCR) | Tesseract | 2β15 MB / language | Apache-2.0 |
| Speech to Text (Transcription) | Whisper base | β 80 MB | MIT |
| AI Translator | OPUS-MT | β 80β110 MB / direction | CC BY 4.0 |
| Background Remover | MODNet | β 26 MB | Apache-2.0 |
| Blur Faces in Photos | MediaPipe BlazeFace | < 1 MB | Apache-2.0 |
| AI Alt Text Generator | Florence-2 base | β 250β450 MB | MIT |
| AI Text Summarizer | Flan-T5 small | β 150 MB | Apache-2.0 |
| AI Writing Assistant | Flan-T5 small | β 150 MB | Apache-2.0 |
| Text to Speech | Device voices | None | β |
Translation models: OPUS-MT by Helsinki-NLP, University of Helsinki (CC BY 4.0).
AI guides
- How to Extract Text from an Image or Scanned PDF (OCR)
- How to Transcribe Audio to Text for Free (and Make Subtitles)
- How to Remove the Background from a Photo
- How to Blur Faces in a Photo Before Sharing It
- How to Write Good Alt Text for Images (with AI Help)
- How to Translate Text Privately Without Sending It Online
- What Is On-Device AI? Private AI Tools That Run in Your Browser
Frequently asked questions
Are my files sent to an AI service?
No. The models run in your browser. Only the model files are downloaded, and they contain no information about your content.
Why is the first use slower?
Your browser downloads and prepares the model. Afterwards it reuses the cached copy.
Do the tools work on phones?
Yes, in modern browsers, although larger models can be slow or run out of memory on older phones.