I’m using it with the llama model. It’s super fast and works pretty well for code at least.