Comment by evanjrowley

Try llama.cpp with the biggest LLM you can find.