Local/Chat - Fully Client-Side AI Chat Application
Fully client-side AI chat app that runs large language models directly in the browser with no backend server or API calls.
Features
Multi-model support
Streaming responses
Offline inference
Local persistence
Network activity ledger
Installable PWA
Responsive UI
Tools
WebLLM
WebGPU
React
Vite
PWA
Project Context
Local/Chat is a browser-first AI chat application that loads open-source models with WebLLM and runs inference entirely on the user's GPU via WebGPU. Once a model is downloaded and cached, the app continues working offline, including model inference, chat history, and settings. The experience is built around privacy-first design, with all conversations and preferences stored locally and no external sync.