Nyvera

Nyvera is a progressive web app (PWA) that lets you chat with AI models powered by WebLLM. Featuring a modern, ChatGPT-like interface, it's designed to look sleek, run fast, and work completely offline. All AI processing happens locally in your browser using WebGPU - no server required, complete privacy guaranteed. Once installed, Nyvera behaves like a native app on desktop or mobile, complete with icons, splash screens, and offline support.

✨ Features

• Modern ChatGPT-like Interface – Clean, professional UI with sidebar navigation, conversation management, and dark theme.

• Completely Offline – All AI processing happens locally in your browser using WebGPU. No internet required after initial setup.

• Conversation Management – Create multiple chat sessions, switch between them, and maintain conversation history locally.

• Sidebar Navigation – Collapsible sidebar with conversation list, model selection, and settings.

• Model Selection – Choose different AI models from WebLLM's extensive library.

• Local AI Inference – Powered by WebLLM for true offline AI chat with complete privacy.

• PWA Support – Installable on desktop and mobile, with offline caching.

• Responsive Design – Optimized for desktop, tablet, and mobile devices.

• Cross-Platform Support – Works on Chrome, Edge, Safari, and mobile browsers.

🚀 Getting Started

1. Clone the Repository

git clone https://github.com/Nyvera/Nyvera.github.io.git
cd Nyvera.github.io

2. Run Locally

You need a local server to test service workers (they don't run from file://).

Quick option: use Python's HTTP server:

# Python 3
python -m http.server 8080

Then visit: http://localhost:8080

3. Deploy

• GitHub Pages: Push to a GitHub repo and enable GitHub Pages in your repo settings.
• Netlify/Vercel: Drag and drop the project folder into your dashboard.
• Custom Hosting: Upload all files to your web server root.

📱 Installation (as PWA)

1. Open Nyvera in Chrome, Edge, or Safari.
2. Click the "Install App" prompt (or "Add to Home Screen" on iOS).
3. Launch it like a normal app — complete with splash screen and custom icon.

⚡ Offline Support

• When online, assets and API calls are cached automatically.
• If offline, cached assets load instantly.
• Models can run completely offline once loaded.
• This ensures the app feels responsive even without an internet connection.

🎨 UI Features

Sidebar Navigation

• Collapsible sidebar for easy access to conversations and settings
• "New Chat" button to start fresh conversations
• Conversation history with easy switching between chats
• Model selection integrated into the sidebar

Chat Interface

• Clean, modern message bubbles for user and AI responses
• Markdown and code syntax highlighting support
• Copy button for AI responses
• Example prompts to get started quickly
• Smooth animations and transitions

Settings

• Centralized settings modal
• Export individual conversations
• Clear all chat history
• Model information display

🛠 Customization

Colors & Theme

The new interface uses a dark theme inspired by ChatGPT and OpenWebUI. You can customize colors by modifying the CSS variables in index.html:

:root {
  --bg-primary: #171717;
  --bg-secondary: #212121;
  --accent: #10a37f;
  /* ... and more */
}

API Integration

Right now, the chat uses WebLLM to chat online and offline. You can swap this out with any other AI API by modifying the JavaScript logic inside index.html.

Caching

Adjust what's cached offline by editing the ASSETS array in sw.js.

🔒 Privacy & Security

This app:

• 100% Private - All AI processing happens locally in your browser using WebGPU
• Does not collect or transmit user data
• Does not require login or account creation
• Works completely offline after initial model download
• Stores conversation history locally in your browser only
• No external API calls for AI inference

🧩 Browser Compatibility

• Chrome / Edge (Desktop & Android) → Full support (PWA install, offline, WebGPU for AI models).
• Safari (iOS & macOS) → Supports install and Apple Touch Icon, but service workers have limited background functionality.
• Firefox → Works as a website but limited PWA features.

Note: WebGPU is required for running AI models. Please use Chrome/Edge 113+ or check if WebGPU is enabled in your browser flags.

📜 License

This project is open-source under the MIT License. You are free to use, modify, and distribute it with attribution.

🙌 Credits

• UI Design: Inspired by ChatGPT
• Built with: Vanilla JavaScript, HTML5, CSS3
• Powered by: WebLLM for offline AI inference
• Syntax Highlighting: Highlight.js
• Markdown Rendering: Marked.js
• Icon design: Nyvera

🌟 What's New

Latest Update: Focused Offline Experience

• Removed image generation to focus on local AI chat
• Enhanced ChatGPT-like styling with better animations
• Improved offline capabilities - truly works without internet
• Better performance with streamlined codebase
• Complete privacy with 100% local AI processing
• Complete UI overhaul with a modern, ChatGPT-like design
• Added sidebar navigation with conversation management
• Implemented conversation history and switching
• Added settings modal for better configuration
• Improved mobile responsiveness
• Enhanced message styling with copy functionality
• Added example prompts for quick start
• Better status indicators and progress tracking