Frankly, I am just trying a bunch of stuff/libraries/demo apps to see where we are at with local LLMs. Most of the stuff I am seeing are just very poor ports of server runtimes, which is terrible.