Hi Huzzlers,

I'm excited to introduce VT Chat – a minimal AI chat application with true local-first architecture and advanced research capabilities.

What makes it different:

Your conversations stay completely local in IndexedDB with zero server storage. Your API keys never leave your browser – I can't see them even as the developer. Switch seamlessly between 15+ AI models including Claude 4 Sonnet/Opus, O3, Gemini 2.5 Pro, and DeepSeek R1 across OpenAI, Anthropic, Google, and other providers.

Key features:

• Deep Research: Multi-step research with source verification
• Pro Search: Real-time web search powered by Google Gemini
• Document Processing: PDF analysis and structured JSON extraction
• Thinking Mode: See complete AI reasoning chains
• Semantic Routing: AI automatically activates tools based on your queries

Built with Next.js 14, TypeScript, and Turborepo. Fully open source for self-hosting. The hosted version keeps core features free, with optional VT+ for premium models and advanced research.

What started as a weekend experiment became a year-long journey through many iterations. I've learned tremendously from the open-source community and builders sharing their work publicly.

Happy to discuss the local-first design decisions, research architecture, or any technical details. Thanks for reading!

https://vtchat.io.vn