Comment by synaesthesisx

How this compare to just running llama.cpp locally?