From the repo:

Run LLMs locally on Cloud Workstations. Uses:

Quantized models from [Huggingface]

llama-cpp-python's webserver

But sure, the blog post doesn't mention it.

Yes, I was talking about this.