🟢 Beginner ⚙️ Type: PWA / Local Browser AI 💸 Free & Open Source ⭐ 15+ GitHub Stars
What is ChromeChat?
ChromeChat is an innovative Progressive Web App (PWA) that allows you to run a powerful AI chatbot directly inside your browser—completely offline, with no server, no cloud subscriptions, and absolutely no API keys required.
Instead of relying on remote servers to process your prompts, ChromeChat taps into the cutting-edge AI models natively built into modern browsers. It utilizes Google’s experimental Prompt API to run Gemini Nano directly within Google Chrome, or Phi-4 if you are using Microsoft Edge.
Because the app runs locally on your hardware, it is exceptionally fast and 100% private. All your chats, images, and document uploads are processed on your own machine and saved safely within your browser’s local storage.
Who is it for?
- Privacy-conscious individuals who want to interact with AI without their personal data, code snippets, or documents ever leaving their computer.
- Travelers and remote workers who need access to a reliable, smart assistant in environments without an active internet or Wi-Fi connection.
- Developers and tech enthusiasts looking to explore Chrome’s native AI capabilities and the experimental Prompt API.
- Anyone looking for a completely free, fast alternative to cloud-based tools like ChatGPT without the heavy technical setup of desktop apps like LM Studio or Ollama.
What makes it special?
- 100% Offline Capability — Simply visit the page once on Wi-Fi to download the model, and you can chat from anywhere in the world completely offline.
- Zero Setup Servers — You don’t need to touch Python, terminal commands, or complicated Docker containers. It works right inside the browser window.
- Rich File Attachments — You can upload PDFs, text files, code snippets, and images for the AI to analyze and summarize natively.
- Secure Local Storage — Your chat history is saved to LocalStorage, while large attachments are stored in IndexedDB. Nothing is ever sent to a third-party server.
- Self-Contained Chat Exports — You can easily import and export your chat logs as
.crchatfiles, which automatically embed your file attachments so you never lose context.
Requirements before you start
Before using ChromeChat, ensure your device meets these system requirements:
- Google Chrome (v148+) or Microsoft Edge — The browser must be modern enough to support the built-in AI LanguageModel APIs. (Note: Chromium forks like Brave may spoof the API and are not recommended).
- ~4 GB of Free Disk Space — Required for the one-time local download of the Gemini Nano model (or ~2.4 GB for Microsoft Edge’s Phi model).
- Compatible Hardware — A modern CPU and GPU are required to process the AI generation smoothly. The app will automatically check your hardware compatibility on launch.
- Wi-Fi Connection — Needed only for the very first visit to download the AI model.
Step-by-step installation
Step 1 — Enable the Chrome AI Flag
Because the built-in AI is an experimental feature, you need to turn it on manually. Open Google Chrome and paste this exact link into your address bar:
chrome://flags/#prompt-api-for-gemini-nanoSet the dropdown menu to Enabled, then click the Relaunch button that appears at the bottom of the screen.
Step 2 — Verify the Model Component
Next, paste this into your Chrome address bar:
chrome://componentsScroll down until you find Optimization Guide On Device Model. If it says “Update out of date”, click the Check for update button to ensure Chrome prepares the model.
Step 3 — Launch the Web App
Navigate to the live ChromeChat application:
https://nay-cat.github.io/ChromeChat/(Optional: You can click the “Install App” icon in your browser’s address bar to save it as a native desktop application window.)
Step 4 — Download the Model
On your first launch, ChromeChat will detect the Prompt API and guide you through the model download. Select your preferred language (English, Spanish, or Japanese) and wait for the ~4 GB download to finish. Keep the tab open during this process.
Step 5 — Go Offline and Chat
Once the download is complete, you are ready to go! You can customize your name, the AI’s name, and the system prompt in the settings. Disconnect from your Wi-Fi to test the true offline privacy of your new browser AI.
Common errors and fixes
| Error | What it means | How to fix it |
|---|---|---|
Model API detected but not responding | The browser flag is enabled, but the actual AI model component failed to download or initialize in the background. | Paste chrome://on-device-internals into your address bar to view the exact download status, file path, and underlying model errors. Restarting Chrome often forces the download to restart. |
Storage Limit Reached / Old Chats Deleted | Your chat text has exceeded the browser’s 5 MB LocalStorage limit. | The app automatically deletes the oldest chats to make room. To prevent data loss, regularly export your important conversations as .crchat files. |
| App stuck loading on Brave or Helium | Alternative browsers often spoof the API features but lack the actual Google/Microsoft AI integration. | Switch to the official Google Chrome or Microsoft Edge browser for guaranteed compatibility. |
Free vs Paid comparison
| Feature | ChromeChat (Free Local AI) | Cloud AI Bots (ChatGPT / Claude) |
|---|---|---|
| Monthly Cost | $0 Forever | $20+ per month for premium features |
| Offline Usage | ✅ Yes — fully functional without Wi-Fi | ❌ No — requires constant internet connection |
| Data Privacy | ✅ 100% Local — data stays in your browser | ❌ Server-side — prompts are logged by corporations |
| Reasoning Capability | 🟡 Moderate — uses smaller, highly optimized models | 🟢 High — uses massive data center server farms |
Bottom line: If you are dealing with sensitive documents, traveling offline, or simply want a private AI assistant without dealing with Python installations, ChromeChat is a brilliant, lightweight solution. However, for extremely complex coding tasks or massive data analysis, a premium cloud model will still outperform the built-in browser nano-models.
Alternatives — 3 similar tools
1. WebLLM
An incredible open-source project that runs standard LLMs (like Llama 3) entirely in your browser using WebGPU. It requires no installation, but the models are downloaded directly into your browser cache and heavily rely on your hardware’s graphics card capabilities.
2. LM Studio
A full desktop application that allows you to easily discover, download, and run any open-source model locally on your PC or Mac. It is more powerful than ChromeChat but requires a dedicated software installation and much more storage space.
3. Jan
An open-source, privacy-first desktop alternative to ChatGPT that runs completely offline. It features a beautiful, clean interface and allows you to swap out different local models with a single click, acting as an excellent stepping stone into local AI.
🔗 jan.ai
🚀 Want more free AI tools like this?
We find, test, and write setup guides for the best free and open-source AI tools — so you don’t have to dig through GitHub yourself.Browse Free AI Tools at globalaiforce.com/shop →
📸 Follow us for daily AI tool tips and tutorials: instagram.com/globalaiforce