Has anyone in the group been playing with Google’s Gemma 4 yet? Of all the recent “hot new releases”, it’s the only one that’s manageable on my modest hardware.
I’ve thrown 26B onto my M4 MacBook Air (using LMStudio 0.4.11) last week and have been testing it out with a few known knowns (e.g., popping a device config into it and asking questions I already know the answer to).
Gemma 4 seems pretty sharp, and “cleverer” than GPT-OSS 20B which I’ve also been messing around a with for the past month or so. I’m yet to try either on more modest hardware at home, but will get around to it after some server hardware changes and a bit of a reconfiguration of the homelab.
Has anyone tried these, the newest Qwen, or the latest from z.ai or MiniMax at home? Looks like we’ve all become spoiled for choice again in the past 6-8 weeks or so!
