I use the local LLM for data cleanup and analysis. I've benchmarked plenty of the top open-weight local models on this machine. So, for now, my local LLM really is limited to grunt work. And because I have lots of data work that depends on the LLM, that's made me use ComfyUI less than I'd like. I don't mean that as a general statement, but at least for my use case, it was really, really bloated.