No models loaded across the mesh.
Deploy a vLLM or Ollama worker on any node and register it as a backend
to see models appear here and at POST /v1/chat/completions.