Open source repositories tagged with #on-device-llm, ranked by health score.
Gemma 4 26B-A4B inference in ~2 GB of RAM on any M-series MacBook
