Dynamic Model Loading with Docker Deployment

[ArijitSinghEDA]

If I use the Dockerfile to deploy a vLLM server, does it only support single model deployment, or can I load models dynamically as well?

The Dockerfile contains only this line as its execution:
ENTRYPOINT ["python", "-m", "vllm.entrypoints.openai.api_server"]

In case it does support loading multiple model loading, how do I use it then?