Overtchat: A cleaner, simpler self-hosted ChatGPT alternative for local models
Built for non-technical users, overtchat ships with search and TTS out of the box.
Developer anitamaxwynnn69, running their own 4x RTX 3090 rig with Qwen3.6 27B, found Open WebUI too rough for non-technical family members. Their solution is overtchat, a new open-source project designed to be a self-hosted ChatGPT-style app that 'just works'. The focus is on a polished UI, simple setup, and minimal moving parts, deliberately avoiding the agentic complexity of LibreChat, LobeChat, or Open WebUI.
The project ships with a single Docker Compose file that includes a pre-tested SearXNG configuration for web search and Kokoro TTS (text-to-speech) requiring no API keys. It's MIT-licensed, has zero telemetry, and is optimized as a Progressive Web App (PWA) for mobile use. The developer acknowledges heavy use of AI in coding but promises it is not 'slop' and is actively maintaining the project. Feedback is welcome via GitHub.
- Overtchat is a self-hosted alternative to Open WebUI, optimized for local models like Qwen3.6 27B on multi-GPU rigs.
- Includes built-in SearXNG web search and Kokoro TTS without needing API keys, all in a single Docker Compose file.
- MIT licensed with no telemetry, mobile PWA support, and a focus on simplicity for non-technical users.
Why It Matters
Brings a polished, zero-config local AI chat experience to households, reducing reliance on cloud services.