Write and rewrite
Draft emails and messages, change tone, shorten or expand text.
On-device AI for Android
Private AI chat that runs offline on your Android phone
Offline AI Chat Private downloads a local language model to your phone. After that, it writes, summarizes, translates and answers questions without sending your prompts to a cloud AI service—even in airplane mode.
On a computer? Scan with your Android phone to open the app on Google Play.
- 50K+ downloads
- Installed by more than 50,000 Android users on Google Play.
- Airplane mode
- Answers are generated by the model on your phone.
- From 468 MB
- Free models from 468 MB to 3.9 GB.
- No account
- Install, choose a model, start chatting.
Real app screens
What the app actually looks like
Every image below is the app's real interface, rendered from the current Android build in automated screenshot tests—no mockups, no marketing overlays.
Answers generated on the phoneAsk the Smart model to turn meeting notes into action items. The reply is produced locally, then you can copy it or keep the chat. 
Start from a promptSummarize, rewrite in a formal or friendly tone, translate, explain, brainstorm, plan a meal or ask about a photo. Sizes before you downloadEach model shows its download size and what it handles—text, images, audio or reasoning. 
Switch models any timeThe active model is marked. Premium unlocks larger and specialised models such as Gemma 4 12B and FLUX.2.
The app interface follows your phone's language; a Russian version of these screens is on the Russian page.
What you can do
Everyday AI help that stays on your phone
Summarize notes and documents PDF: Premium
Paste notes or attach TXT and Markdown files. Questions about text-based PDFs are part of Premium.
Translate
Translate typed text both ways and ask for a back-translation to check the meaning.
Ask about a photo
Quick and Smart accept image input, for example a label, a sign or a whiteboard you photographed.
Study and explain
Get plain-language explanations, recall questions and short study plans.
Keep and export chats Export: Premium
Saved chat history is free. Premium exports a conversation as a Markdown file.
Local AI models
Choose a model that fits your phone
Models are downloaded once and run on the device. Bigger models answer better but need more storage and memory; the app checks your phone before a download.
| Model | Download | Based on | Good for |
|---|---|---|---|
| Tiny v2 | 468 MB | Granite 4.0 350M | Very low-memory phones and simple tasks; answer quality is limited |
| Bonsai 1.7B | 766 MB | Qwen3-based Ternary Bonsai | Fast compact chat on phones with a compatible mobile GPU |
| Quick | 2.6 GB | Gemma 4 E2B | Everyday chat, writing and translation; image and audio input |
| Smart | 3.7 GB | Gemma 4 E4B | Harder prompts and longer sessions; image and audio input |
| Compact | 3.9 GB | Phi-4 mini | Text-first, CPU-friendly structured tasks |
Premium adds Gemma 4 12B (6.9 GB), Qwen2-VL-2B for images and OCR, DeepSeek R1 1.5B for step-by-step reasoning, Qwen2.5 1.5B, Qwen2.5 Coder 3B, Qwen3 0.6B and FLUX.2 on-device image generation where supported. Sizes are approximate.
Which model fits your phone?
A practical starting point, not a compatibility promise—the app checks your device before download.
Get it on Google PlayLeave free-space headroom: a model's download size is not all the storage your phone needs to download and run it.
See what local AI can help with
These are prepared examples, not live model responses.
Input: Meeting notes with three decisions and two deadlines. Example output: A five-bullet brief with owners, deadlines, and open questions.
Privacy boundary
What stays on your phone—and what still uses the internet
Local inference means the model runs on your device. Here is exactly where that boundary is.
Stays on your phone
- Your prompts and the model's answers, after the model is downloaded
- Saved chat history
- Attached documents and photos you ask about
- The downloaded model itself
Still needs a connection
- Installing and updating the app
- Downloading or updating models
- Google Play purchases and restores
- Optional analytics and crash reports—only if you allow them
| Typical cloud AI app | Offline AI Chat Private | |
|---|---|---|
| Where answers are generated | On the provider's servers | On your phone |
| Works without internet | No | Yes, once a model is downloaded |
| Account required | Usually | No |
| Your prompts leave the phone | Yes, sent to the provider | No |
| Answer quality on hard tasks | Higher: very large models | Good for everyday tasks; limited by the phone |
| Storage on the phone | Small app | Model download from 468 MB to 3.9 GB |
How it works
Download once, then use AI anywhere
Choose
Let the device check recommend a model, or pick one yourself.
Download
Download the model over Wi-Fi and keep extra free storage for a smooth start.
Work offline
Chat, write, summarize and translate—also in airplane mode.
Guides
Useful answers before you download
Which model to downloadTiny, Quick, Smart or Compact: size, RAM and speed.Offline AI explainedWhere a local model runs and what “offline” really means.Private AI on your phoneLocal processing and the honest internet boundary.Local AI vs cloud AIPrivacy, quality, speed and battery compared.Everyday offline AI tasksWriting, summaries, translation, photos and study.Offline AI for travelPrepare a helpful assistant before the signal disappears.
FAQ
Before you install
Does Offline AI Chat Private work without internet?
Yes. After a model is downloaded, AI inference works offline. Internet is needed for installation, updates, model downloads, and Google Play purchases or restores.
Are my prompts sent to a server?
No cloud AI inference is used. Prompts, answers and attachments are processed on your device, and chat history is saved on the phone.
Which phones are supported?
Android 10 or newer. Tiny v2 suits low-memory phones; Quick, Smart and Compact run best with 8 GB of RAM or more and several gigabytes of free storage. The app checks your device before a download.
How big are the AI models?
The free models range from about 468 MB (Tiny v2) to 3.9 GB (Compact). Quick is about 2.6 GB and Smart about 3.7 GB. Keep extra free space beyond the download size.
Can I work with documents?
Attach TXT or Markdown documents for free. Premium adds questions about text-based PDFs; scanned and password-protected PDFs are not supported.
Is the app free?
Yes. The app, the free models and saved chat history are free. Premium is an optional in-app purchase that adds more models, PDF questions and Markdown export.
Bring private, offline AI to your Android phone
Download a local model once, then get help writing, summarizing, translating, studying and working with text—wherever you are.