Comment by ixaxaar

I'm running it using a 4090 on using llama.cpp with Q5_K_S and its running at ~33 t/s