Faster Chat

Faster Chat is an open source chat interface for large language models that runs on the user's own machine or server. Conversations are stored locally in the browser through IndexedDB, with optional SQLite sync, so chat history never has to leave the device. It connects to local inference servers such as Ollama or LM Studio for fully offline use, or to hosted providers including OpenAI, Anthropic, Groq and Mistral when a stronger model is needed. It is aimed at developers who want the capability of a commercial chat product without handing over their data or depending on a vendor's uptime. An admin panel handles provider connections, discovers available models automatically through a model directory integration, and accepts any compatible API endpoint as a custom provider. Accounts support multiple users with role based access control across administrator, member and read only roles, plus a user management panel. The interface renders markdown, highlights code and streams responses, using a stack chosen for speed: Preact, Hono, Tailwind, the Vercel AI SDK, Dexie for local persistence and TanStack for typed routing. Deployment is a single Docker command with optional HTTPS, or a local run with Bun or Node.js version 20 and above. The project is free under the MIT license. There is no pricing page, hosted plan or paid edition. Running costs are whatever the chosen model provider charges, and nothing at all when pointing it at a local model. The site lists version 0.2.0 as the current release, so this is still early software and worth evaluating against that.