I think they were asking, "If this is a model produced by Meta with an API designed to be compatible with OpenAI models, but it's not actually an OpenAI model, why is Meta hosting it on Azure?"

[deleted]

GPU compute capacity?