Connect an AI model
angelexe doesn't ship its own model — it talks to whatever OpenAI-compatible endpoint you point it at. This works with self-hosted open-source models or a hosted provider that uses the same API shape.
Heads up on going public: settings are stored per-browser, in each visitor's own
localStorage — nothing here is shared between people. That means if you deploy this for other
people to use, everyone configures (and pays for) their own AI connection. There's no
shared server-side key, and no way to bill or rate-limit usage without adding a real backend.
For a personal setup this is perfectly fine. If you want a public product where you supply the
AI for everyone, you'll need a small server (even a free-tier one) to hold the API key and proxy
requests — happy to help you build that next.
Popular open-source options
- Ollama — easiest way to run models like Llama 3 or Mistral locally, exposes an OpenAI-compatible endpoint out of the box.
- text-generation-webui — more configurable local UI + API for open models.
- LM Studio — desktop app, one click to spin up a local OpenAI-compatible server.
- Any hosted provider serving open-weight models over an OpenAI-style API also works — just fill in their URL, model name, and key.