VELLUR

AI Core

Private OpenAI-compatible gateway. Inference runs on a separate GPU workspace that is switched on only when needed.

Gatewaychecking…

Endpoint

Modelsenter a key below

Inference backendenter a key below

Gateway key (kept only in this browser tab)

The test prompt uses the GPU for a second. Nothing here is simulated: if no backend is connected, you will see the gateway's real error.

Administration happens on the server only (sudo vac). This page has no admin functions.