{ ... }: {
  den.aspects.llama = {
    nixos = { pkgs, ... }: {
      services.llama-cpp = {
        enable = true;
        package = pkgs.llama-cpp.override { cudaSupport = true; };
        settings = {
          host = "0.0.0.0";
          port = 8080;
          ctx-size = 65536;

          models-preset = (pkgs.formats.ini { }).generate "llama-cpp-models-preset.ini" {
            "Qwen3.6-35B-A3B" = {
              hf-repo = "unsloth/Qwen3.6-35B-A3B-GGUF";
              hf-file = "Qwen3.6-35B-A3B-UD-Q4_K_XL.gguf";
              alias = "qwen3.6-35b-a3b";
              temp = "1.0";
              top-p = "0.95";
              top-k = "20";
              n-gpu-layers = 80;
              ctx-size = 65536;
            };
          };
        };
      };
    };
  };
}