Post

I don't have _any_ uses for an LLM where I can't specify the model it will use. Not until the general standard for mini/on-device models is way better than the current bleeding edge.