wtf is open-poe-ai?

Open-Poe-AI lets you chat with multiple AI models, like GPT, Claude, Gemini, Llama, and others, all from one chat interface that you run on your own server. Think of it like Poe.com, but self-hosted: you bring your own API keys from whichever AI providers you want to use, and you keep full control over your conversations and data instead of routing everything through a third-party platform. The project gives you a single web-based chat UI where you can switch between models, query several at once in the same thread to compare answers, and even build custom bots with their own instructions and knowledge. It supports text, image generation, vision, and audio. You can also have group chats where multiple users and multiple AI models participate together. Setting it up involves copying a configuration file, adding your API keys, and running a single deployment command, after which you access the interface through your browser. This is for someone who wants the convenience of a unified AI chat tool but doesn't want to depend on a hosted service. A startup team could use it internally to compare model outputs side by side without each person needing separate accounts. A hobbyist running local models could connect those alongside commercial APIs. Anyone concerned about data privacy, where conversations shouldn't pass through someone else's servers, gets that control by hosting it themselves. The project is explicitly early-stage and calls for contributions. It's part of a broader collection of open-source generative AI apps by the same author. The README doesn't go into much detail on the technical architecture beyond mentioning that it supports any OpenAI-compatible endpoint, including locally hosted models, and that deployment is handled through Docker Compose with support for Kubernetes.