I tried Mixtral on my laptop but was only able to get about 6 tokens per minute. Still pretty incredible that it can run on high-end consumer hardware! Open source AI is catching up quick.