Llama.cpp

{
  "models": [
    {
      "title": "<MODEL_NAME>",
      "provider": "llama.cpp",
      "model": "<MODEL_ID>"
      "apiBase": "http://localhost:8080"
    }
  ]
}