{
  "slug": "vllm",
  "name": "vLLM",
  "description": "High-throughput, low-latency LLM serving engine with PagedAttention for production workloads.",
  "category": "AI & Voice",
  "image": "vllm/vllm-openai:latest",
  "logo": "https://cdn.jsdelivr.net/gh/homarr-labs/dashboard-icons/png/vllm.png",
  "website": "https://docs.vllm.ai",
  "docsUrl": "https://docs.vllm.ai/en/latest/serving/deploying_with_docker.html",
  "ports": [
    {
      "host": 8000,
      "container": 8000,
      "protocol": "tcp"
    }
  ],
  "volumes": [
    {
      "host": "/opt/appdata/vllm/models",
      "container": "/root/.cache/huggingface"
    }
  ],
  "envVars": [
    {
      "key": "TZ",
      "default": "Etc/UTC",
      "description": "Timezone"
    },
    {
      "key": "HF_TOKEN",
      "default": "",
      "description": "Hugging Face token for gated models",
      "secret": true
    }
  ],
  "minMemory": "8GB",
  "version": "latest"
}