Hacker News new | past | comments | ask | show | jobs | submit
kadoban | on: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
Is the quality of the thinking better or it's just shorter since Mandarin is more compact?