🌌 Aurora — 100% Free AI Chat, Runs in Your Browser

No API keys. No signups. No servers. No hosting bill. Ever.

---

How it works now

The AI model runs entirely inside the visitor's browser using WebLLM + WebGPU. There is no backend, no server, nothing to deploy except plain static files.

• First visit: browser downloads a small model (~700MB, one time, then cached)
• Every message after that: answered instantly, 100% on-device
• Zero cost for you, zero cost for anyone who visits — it's their device doing the work

This replaces the old server/API-key/hosting approach completely.

---

Deploy for free — GitHub Pages

1. Push this whole client/ folder's built output to a gh-pages branch, or use GitHub's built-in Pages workflow:

   cd client
   npm install
   npm run build

   This creates a dist/ folder — that's your entire website.

2. In your GitHub repo: Settings → Pages → Deploy from a branch, and either:

   • push the contents of dist/ to a gh-pages branch, or
   • use a GitHub Action (recommended) — add .github/workflows/deploy.yml:

   name: Deploy to GitHub Pages
   on:
     push:
       branches: [main]
   jobs:
     build-and-deploy:
       runs-on: ubuntu-latest
       steps:
         - uses: actions/checkout@v4
         - uses: actions/setup-node@v4
           with:
             node-version: 20
         - run: cd client && npm install && npm run build
         - uses: peaceiris/actions-gh-pages@v3
           with:
             github_token: ${{ secrets.GITHUB_TOKEN }}
             publish_dir: ./client/dist

3. That's it. Your site is live at https://<username>.github.io/<repo>/, free forever, no server anywhere.

---

Run locally first (optional)

cd client
npm install
npm run dev

Open the URL Vite prints. Needs a WebGPU-capable browser (recent Chrome/Edge).

---

Notes

• Requires a browser with WebGPU (Chrome/Edge 113+, most 2023+ browsers). No WebGPU = no chat, since there's no server fallback.
• The model (Llama-3.2-1B-Instruct) is small and fast but not as capable as large cloud models — that trade-off is what makes it free and serverless.
• To use a bigger/smaller model, change MODEL in client/script.js to any name from the WebLLM model list.

---

👨‍💻 Author

Anusthan Singh

• GitHub: https://github.com/anusthan12
• LinkedIn: https://www.linkedin.com/in/anusthan12/

📄 License

MIT License © 2023–2026 Anusthan Singh