Self-Hostable AI Chat Application

A modern, feature-rich AI chat application built with Svelte 5, Bun, and OpenRouter. Performance-optimized, secure, and supports multi-model conversations with intelligent auto-routing, Ollama integration, and comprehensive analytics.

Table of Contents

• Features
• Quick Start
• Docker Deployment
• Desktop App (Electron)
• Mobile App (Android via Capacitor)
• Project Structure
• Configuration
• Usage
• Development
• Security Considerations
• Roadmap
• Contributing
• License
• Acknowledgments
• Support

✨ Features

🤖 AI Capabilities

• Auto Mode: Intelligent model routing using arcee-ai/trinity-mini:free to select the best model for each query.
• Manual Multi-Model Mode: Run multiple models simultaneously and compare responses side-by-side.
• Ollama Integration: Support for local models via Ollama.
• Zero Data Retention (ZDR): Toggle to use only privacy-compliant models with ZDR policies.
• Real-time Streaming: Token-by-token streaming with smooth animations and Svelte 5 reactivity.
• Chat Summaries: Automatic generation of descriptive chat titles.
• Image Generation: Native image generation with Gemini, Seedream, and GPT-5 models.
• Web Search: Integration with citations for up-to-date information.

📊 Analytics & Stats

• Per-message statistics (tokens, cost, latency).
• Multi-model aggregated stats.
• Dashboard with time-series charts (cost, usage, performance).
• Cost tracking and optimization insights.

🔐 Security & Privacy

• AES-256-GCM encrypted API key storage.
• Optional client-side encryption with IndexedDB.
• Data Management: Quickly clear local cache or database records.
• Secure, self-hosted session management.

🎨 User Experience

• Performance: Virtualized sidebar list for handling thousands of chat histories smoothly.
• Modern UI: Dark developer-tool aesthetic using Tailwind CSS v4.
• Responsive: Three-column layout with collapsible sidebars.
• Cross-Platform: Web, Desktop (Electron), and Mobile (Android).

🚀 Quick Start

Prerequisites

• Bun runtime (required)
• OpenRouter API key (Get one here)
• (Optional) Ollama for local models

Installation

1. Clone the repository

git clone https://github.com/aydenstechdungeon/selfhostablechat
cd selfhostablechat

2. Install dependencies

bun install

3. Set up environment variables

cp .env.example .env

4. Start the development server

bun run dev

Visit http://localhost:5173

🐳 Docker Deployment

The easiest way to self-host this application is using Docker.

Quick Deploy

1. Build and run with Docker Compose

docker-compose up -d

2. Access the application

http://localhost:3420

Docker Commands

# View logs
docker-compose logs -f

# Rebuild after updates
docker-compose up -d --build

# Stop the application
docker-compose down

🖥️ Desktop App (Electron)

Build a native desktop application for Linux, Windows, and macOS.

# Development
bun run desktop

# Build (Linux AppImage)
bun run electron:build:linux

# Build (Windows EXE)
bun run electron:build:win

# Build (macOS DMG)
bun run electron:build:mac

Built applications will be in dist-electron/.

📱 Mobile App (Android via Capacitor)

Build a native Android application with a bundled Node.js server.

# Build SvelteKit and sync Capacitor
bun run mobile

# Open in Android Studio
bun run mobile:open

See CAPACITOR_IMPLEMENTATION.md for details.

📁 Project Structure

src/
├── lib/
│   ├── components/      # Svelte 5 components (Runes, Snippets)
│   ├── server/          # AI routing, streaming, and database
│   ├── stores/          # Reactive stores for state
│   ├── shared/          # Shared constants (Model definitions)
│   └── types.ts         # TypeScript definitions
├── routes/
│   ├── api/             # SSE streaming and analytics endpoints
│   └── (app)/           # Main UI routes (Chat, Dashboard, Settings)
static/                  # Assets (Icons, WebP images)

🔧 Configuration

Supported Models (Auto-Routing)

• arcee-ai/trinity-mini:free - Primary Router/Summarizer
• google/gemini-3-flash-preview - Latest Gemini with vision
• google/gemini-3.1-pro-preview - High-quality reasoning
• anthropic/claude-sonnet-4.6 - Advanced coding & writing
• anthropic/claude-opus-4.6 - Complex reasoning
• deepseek/deepseek-r1-0528:free - Deep reasoning
• openai/gpt-4o-mini - Fast, efficient
• x-ai/grok-4.1-fast - Fast general purpose

Local Models (Ollama)

The application can connect to a local Ollama instance (default: http://localhost:11434). Configure this in the Settings panel or via environment variables in future updates.

🔒 Security Considerations

• API Keys: Stored with AES-256-GCM encryption. Use server-side storage for the best security.
• Privacy: Zero Data Retention mode filters for providers that guarantee your data isn't used for training.
• Self-Hosting: You own your database (SQLite) and data.

🎯 Roadmap

• Core chat functionality with Svelte 5 Runes
• Auto-routing with Trinity Mini
• Ollama local model integration
• Multi-model comparison mode
• Desktop & Mobile (Android) app builds
• Image Generation (Gemini, Seedream)
• Web search integration
• Virtualized sidebar for huge chat histories
• Data management features (Clear Cache)
• iOS app support (Capacitor)
• Voice input/output support
• Advanced plugin system

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📄 License

MIT License - feel free to use this project for personal or commercial purposes.

---

Happy chatting! 🚀