Hi, maybe there should be build for two version. Not everyone want's to run local llm models. 