services:
  # 1. vLLM 服务（OpenAI 兼容 API）
  vllm:
    image: vllm/vllm-openai:v0.18.0-cu130
    runtime: nvidia
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    ports:
      - "18000:8000"
    volumes:
      - ./vllm-cache:/root/.cache/huggingface
    environment:
      - CUDA_VISIBLE_DEVICES=0
      - HF_ENDPOINT=https://hf-mirror.com
    command: |
      --model Qwen/Qwen2.5-7B-Instruct-AWQ
      --quantization awq
      --dtype auto
      --gpu-memory-utilization 0.9
      --max-model-len 8192
    restart: always

  # 2. OpenClaw（官方 ghcr 镜像，gateway + cli 合一）
  openclaw:
    image: ghcr.io/openclaw/openclaw:latest  # 官方正确镜像
    ports:
      - "18789:18789"
    environment:
      # 对接 vLLM（OpenAI 兼容）
      - OPENCLAW_LLM_PROVIDER=openai
      - OPENCLAW_LLM_BASE_URL=http://vllm:8000/v1
      - OPENCLAW_LLM_API_KEY=sk-123456
      - OPENCLAW_LLM_MODEL=Qwen/Qwen2.5-7B-Instruct-AWQ
      # 官方配置
      - OPENCLAW_GATEWAY_PORT=18789
      - OPENCLAW_GATEWAY_BIND=lan
    volumes:
      - ./openclaw:/home/node/.openclaw:rw  # 官方挂载路径
    user: "1000:1000"  # 容器内非 root 用户，匹配宿主机目录权限
    restart: always
    depends_on:
      - vllm

  # 3. Open WebUI（常用界面）
  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    ports:
      - "18002:8080"
    environment:
      - OPENAI_API_BASE_URL=http://vllm:8000/v1
      - OPENAI_API_KEY=sk-123456
      - HF_ENDPOINT=https://hf-mirror.com
    volumes:
      - ./open-webui:/app/backend/data
    restart: always
    depends_on:
      - vllm

  # 4. ChatGPT Next Web（简洁界面）
  chat-web:
    image: yidadaa/chatgpt-next-web:v2.16.1
    ports:
      - "18001:3000"
    environment:
      - BASE_URL=http://vllm:8000/v1
      - API_KEY=sk-123456
      - DEFAULT_MODEL=Qwen/Qwen2.5-7B-Instruct-AWQ
    restart: always
    depends_on:
      - vllm