Definitely taking some getting used to being local AI only.
Shit is much slower than cloud, so gotta basically write huge plans then let the clankers bang them out overnight.
Can't do that thing whe...
I’m running Qwen3.8-27b on my MacBook Pro M4 Pro with 48 GB RAM. It definitely takes its time, but recent oMLX updates have made the dense model running a lot quicker than times past. I could only run MoE models before due to speed.
❤️1