services:
  llama-cpp:
    image: ghcr.io/ggml-org/llama.cpp:server-cuda
    container_name: llama-cpp
    restart: unless-stopped
    networks:
      - traefik
    volumes:
      - /pwspool/software/llama-cpp/models:/models
    # We moved your environment variables here to guarantee they are applied
    command: 
      - "--model"
      - "/models/Qwen3.5-35B-A3B-UD-IQ2_XXS.gguf"
      - "--host"
      - "0.0.0.0"
      - "--port"
      - "8080"
      - "--n-gpu-layers"
      - "99"
      - "--ctx-size"
      - "8192"
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    labels:
      - "traefik.enable=true"
      - "traefik.http.routers.llama.rule=Host(`llm.whitney.rip`)"
      - "traefik.http.routers.llama.entrypoints=websecure"
      - "traefik.http.routers.llama.tls.certresolver=lets-encrypt"
      - "traefik.http.services.llama.loadbalancer.server.port=8080"

networks:
  traefik:
    external: true