hey guys, we've optimised our inference engine to run Meta Muse Glimmer 30B faster than any other engine (llama.cpp etc) on Mac. Best performance on M5 Pro.