This is a small door into a local language model running at home — not a cloud chat product. Ask questions in the browser, or give an app a key and talk to the same models over a familiar API.
Models live on a Mac on the network. This site is the friendly front door: it checks your key, then forwards your request to the local engine.
Open the chat page, paste a key, pick a model, and write normally. Replies come from software on this host — not from a big public AI service.
Cursor, scripts, and other clients can call /v1/chat/completions with a Bearer sk-… key, the same way they talk to OpenAI-shaped APIs.
Create and revoke keys from API keys. Without a key, the models stay private. With one, you can use chat or the API.
Same host as the chat UI. Create a key first, then paste it below.
# Say hello through the OpenAI-compatible endpoint curl -s https://ai.suherman.net/v1/chat/completions \ -H "Authorization: Bearer sk-…" \ -H "Content-Type: application/json" \ -d '{"model":"qwen2.5-coder:7b","messages":[{"role":"user","content":"hi"}]}'