⚡ Faster Chat

A blazingly fast, privacy-first chat interface for AI that works with any LLM provider—cloud or completely offline.

Connect to OpenAI, Anthropic, Google, Groq, Mistral, xAI, DeepSeek, and more—or run completely offline with Ollama, LM Studio, or llama.cpp. Your conversations stay on your machine. No vendor lock-in, no tracking, full control.

✨ Features

Core

• 💬 Real-time streaming chat with Vercel AI SDK
• ⚡ Blazingly fast — 3KB Preact runtime, zero SSR overhead, instant responses
• 🗄️ Server-side SQLite storage — Conversations persist across devices and browser tabs
• 🤖 17 built-in providers (plus 50+ more auto-discoverable via models.dev): OpenAI, Anthropic, Google, Google Vertex, Amazon Bedrock, Azure, Groq, Mistral, xAI, DeepSeek, Cohere, Fireworks, Cerebras, OpenRouter, Replicate, Ollama, LM Studio
• 🧠 Cross-conversation memory — AI remembers your preferences, projects, and context across chats
• 🔍 Web search — AI can search the web and cite sources inline (Brave Search)
• 🖼️ Image support — Upload images for vision analysis, generate images with DALL-E, FLUX, and OpenRouter models
• 📥 Import conversations from ChatGPT exports (more formats coming soon)
• 📎 File attachments with preview and download
• 📝 Markdown rendering with syntax highlighting (Shiki) and LaTeX support
• 🎨 Themable UI — 15+ color themes, dark/light mode, custom fonts, syntax highlighting themes
• 🎤 Voice input/output — Speech-to-text and text-to-speech capabilities
• ⌨️ Keyboard shortcuts for power users (Ctrl+B sidebar, Ctrl+Shift+O new chat, etc.)
• 📱 Responsive design for desktop, tablet, and mobile

Administration

• 🔐 Multi-user authentication with role-based access (admin/member/readonly)
• 🔌 Provider Hub: Auto-discover models with models.dev integration
• ⬇️ Pull Ollama models directly from Admin Panel with progress streaming (no CLI needed)
• 🛡️ Admin panel for user management (CRUD, password reset, role changes)
• 🔑 Encrypted API key storage with server-side encryption
• 🎭 White labeling — Customize app name and logo icon for your organization

Deployment

• 🌐 Works completely offline with local models (Ollama, LM Studio, etc.)
• 🐳 One-command Docker deployment with optional HTTPS via Caddy
• 🎨 Modern stack: Preact + Hono + TanStack + Tailwind 4.1

🧠 Memory System

Faster Chat can learn about you across conversations — your preferences, projects, tech stack, and communication style. Memories are extracted automatically after each response and injected naturally into future chats.

• Three-level control: Global toggle (admin), per-user toggle, per-chat opt-out
• Zero latency impact: Extraction happens asynchronously after the response streams
• Full transparency: View, delete individual facts, or clear all memories from Settings
• Configurable extraction model: Use a cheap/fast model (e.g., Haiku, GPT-4o-mini) to minimize cost
• Privacy: Memories are strictly user-scoped. No admin access to user memories.

🔍 Web Search

When enabled, the AI can autonomously search the web and fetch pages to answer questions with up-to-date information.

• Brave Search integration with encrypted API key storage
• SSRF-protected URL fetching with DNS validation and private IP blocking
• Source citations displayed as clickable pills with favicons below the response
• Real-time status: "Searching the web..." and "Reading {domain}" indicators during tool execution
• 5-minute cache for repeated queries, max 5 results per search

🖼️ Image Support

Vision/Multimodal: Upload images alongside messages for analysis by vision-capable models (Claude, GPT-4, Gemini, Grok, Mistral, and more).

Image Generation: Toggle image mode to generate images with configurable aspect ratios (1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3).

• Providers: OpenAI (DALL-E 3), Replicate (FLUX 1.1 Pro), OpenRouter image models
• Generated images display inline with download controls and metadata

🚀 Quick Start

One-Click Docker Deploy (Recommended)

git clone https://github.com/1337hero/faster-chat.git
cd faster-chat
docker compose up -d

That's it. Open http://localhost:8787, register your first user (becomes admin), and configure your AI providers.

With HTTPS (for production):

docker compose -f docker-compose.yml -f docker-compose.caddy.yml up -d

Local Development

Prerequisites: Bun

Note

: This is a Bun workspace using the workspace: protocol, which npm does not support — npm install will fail with EUNSUPPORTEDPROTOCOL. Use Bun for development. To run without Bun installed, use Docker.

git clone https://github.com/1337hero/faster-chat.git
cd faster-chat
bun install
bun run dev

On first run, the server automatically generates encryption keys and initializes the database.

• Frontend: http://localhost:3000
• API Server: http://localhost:3001

Important: Backup server/.env — contains the encryption key for stored API keys.

First-Time Setup

1. Register an account at http://localhost:3000/login
   • The first account is automatically promoted to admin
2. Configure AI providers in the Admin Panel (/admin → Providers tab):
   • Add OpenAI, Anthropic, or other cloud providers with API keys
   • Configure local providers (Ollama, LM Studio) with custom endpoints
   • API keys are encrypted and stored securely server-side
3. Enable models in the Admin Panel (Providers tab → Refresh Models)
   • Select which models appear in the chat interface
   • Set default model for new chats

Configure providers and API keys in the Admin Panel

Enable and manage models from all your providers

New Focus Mode

New Appearance Options, to change colors and fonts

You can now white label and customize the app title and icon

Using Offline with Ollama

# Install Ollama (macOS/Linux)
curl -fsSL https://ollama.ai/install.sh | sh

# In Faster Chat: Admin Panel → Connections → Search "Ollama" → Add
# Then: Admin Panel → Models → Click "Pull Model" on Ollama row → Enter model name

You can pull models directly from the Admin Panel—no CLI needed! Just click Pull Model next to your Ollama provider, enter a model name (e.g., llama3.2, mistral, codellama), and watch the download progress in real-time.

The Provider Hub auto-discovers 50+ providers including Ollama, LM Studio, OpenAI, Anthropic, Groq, Mistral, OpenRouter, and more. Just search and add.

💻 Development

Commands

Root (recommended)

bun run dev         # Start frontend + backend concurrently
bun run build       # Build all packages for production
bun run start       # Run production builds
bun run clean       # Remove all build artifacts
bun run format      # Format code with Prettier

Frontend

cd frontend
bun run dev         # Vite dev server on :3000
bun run build       # Production build to dist/
bun run preview     # Preview production build

Backend

cd server
bun run dev         # Hono dev server on :3001
bun run build       # Build for production
bun run start       # Run production server on :3001

🐳 Docker Details

The Docker setup uses a hybrid build (Bun for deps, Node.js 22 runtime) with SQLite storage in a persistent volume.

HTTPS with Caddy: For production with automatic Let's Encrypt certificates:

docker compose -f docker-compose.yml -f docker-compose.caddy.yml up -d
# Edit Caddyfile with your domain, point DNS, restart

See docs/caddy-https-setup.md and docs/docker-setup.md for details.

Configuration

Environment Variables (server/.env, optional):

server/.env is not required to start. On first run the server generates an encryption key and keeps a copy in the persistent data volume, so it survives container rebuilds. Create the file only to override defaults:

# Encryption key for API keys (auto-generated if omitted)
API_KEY_ENCRYPTION_KEY=...  # Generate with crypto.randomBytes(32)

# Optional: Configure via Admin Panel instead
APP_PORT=8787              # Internal port (default: 8787)
NODE_ENV=production        # Environment mode
DATABASE_URL=sqlite:///app/server/data/chat.db

# For local Ollama access from Docker
OLLAMA_BASE_URL=http://host.docker.internal:11434

Common Commands:

docker compose up -d                # Start
docker compose logs -f              # View logs
docker compose down                 # Stop
docker compose up -d --build        # Rebuild

# Reset database
docker compose down
docker volume rm faster-chat_chat-data
docker compose up -d

🗺️ Roadmap

Faster Chat has reached its stated goal — a lighter, faster, more private alternative to Open WebUI — and is now in maintain mode: quality and polish over new surface area. Everything in the feature list above is shipped and working; what remains is a short list of small, finishable items, each tracked as a PRD in the open issues.

What's next:

• Chat export to JSON/Markdown (#27) and Claude import (#28) — closing the data-ownership loop
• Inline message editing (#29), full-text message search (#30), per-chat system prompt/temperature (#31)
• Minimal personal prompt templates (#34)

Explicitly not planned: knowledge bases/RAG, conversation branching, model arenas, MCP, plugin systems, PostgreSQL, mobile apps, i18n, public share links. Each either grows a second product inside this one or is a permanent maintenance tax. A finished tool that stays sharp beats an ambitious one that rusts — don't become Open WebUI.

🎨 Design Philosophy

Faster Chat is built on these principles:

• Self-Hosted: Your data stays on your server. No cloud dependencies.
• Provider-Agnostic: Never locked into a single AI vendor.
• Minimal Runtime: 3KB Preact, no SSR overhead, instant responses.
• Offline-Capable: Run completely offline with local models.
• Fast Iteration: Bun for speed, no TypeScript ceremony, clear patterns.
• Simple Code: Small focused components, derive state in render, delete aggressively.

Why No TypeScript?

We chose speed over ceremony. TypeScript's compile step and constant type churn across fast-moving AI SDKs slowed development more than it helped.

Our guardrails:

• Runtime validation at system boundaries
• Shared constants and clear contracts
• Tests for critical paths
• JSDoc for complex functions

Trade-off: Less friction, faster iteration, easier contribution.

See WIKI for detailed coding principles and architecture documentation.

🙏 Credits & Acknowledgments

Faster Chat is built on the shoulders of excellent open source projects:

Core Infrastructure

• Vercel AI SDK — Streaming chat completions and multi-provider support
• models.dev — Community-maintained AI model database for auto-discovery
• Preact — Lightweight 3KB React alternative
• Hono — Ultrafast web framework for the backend
• TanStack Router & TanStack Query — Modern routing and server state management
• bun:sqlite — Fast SQLite driver for server-side persistence

UI & Styling

• Tailwind CSS — Utility-first CSS framework
• lucide-preact — Beautiful icon library
• Catppuccin — Soothing pastel theme

External API Calls

For transparency, this application makes the following external API calls:

• models.dev/api.json — Fetches provider and model metadata on server startup (cached for 1 hour)
• Brave Search API — Only when web search is enabled and triggered by the AI during a conversation
• Your configured AI providers (OpenAI, Anthropic, etc.) — Only when you send chat messages
• No tracking, analytics, or telemetry services — Your privacy is paramount

All data (conversations, memories, settings, API keys) is stored in your self-hosted SQLite database. Nothing leaves your server except API calls to your configured AI providers and optional web search.

🤝 Contributing

Contributions welcome! We're looking for:

• Bug fixes and error handling
• New provider integrations
• Documentation improvements
• UI/UX enhancements
• Tests and quality improvements

Before submitting:

1. Read Documentation for coding philosophy and patterns
2. Ensure changes align with our lightweight, offline-first approach
3. Test locally with bun run dev
4. Keep PRs focused on a single feature or fix

📄 License

MIT License — see LICENSE for details.

---

⭐ Star History

If Faster Chat helps you take control of your AI conversations, consider giving us a star!

---

Built with ❤️ by 1337Hero for developers who value privacy, speed, and control.
_{No tracking. No analytics. Just fast, local-first AI conversations.}