Hacker News new | past | comments | ask | show | jobs | submit
dist-epoch | on: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
Gemma4-E4B (4B params) works pretty well as a local wiki, or when you don't have connectivity.
dannyw | parent
Nitpick: Gemma4-E4B is actually a 8 billion param model, but only 4.5B params worth of memory bandwidth needed per decode.