Possibly because it's very NVIDIA focused?

I'm excited about the prospect of small models e.g. fine-tuning against MCP calls to give APIs language interfaces.