home_stack/ai_stack/vllm/docker-compose.cpu.yaml

services:
  vllm:
    image: docker.io/vllm/vllm-openai:latest
    command:
      - --model
      # - Qwen/Qwen3-4B-Instruct-2507
      - Qwen/Qwen3-VL-2B-Instruct
      - --device
      - cpu
      - --max-model-len
      - "2048"
    ports:
      - "9000:8000"
    environment:
      HF_TOKEN: ${HF_TOKEN}
      VLLM_USE_V1: "0"
    volumes:
      - ${HOME}/.cache/huggingface:/root/.cache/huggingface
    ipc: host
    deploy:
      resources:
        reservations:
          # WARN: requires Docker Compose 2.24.2
          # https://docs.docker.com/reference/compose-file/merge/#replace-value
          devices: !override
            - capabilities: ["gpu"]
              driver: cdi
              device_ids:
                - nvidia.com/gpu=all