Comment by xeckr

Looks like it punches way above its weight(s).

How far are we from running a GPT-3/GPT-4 level LLM on regular consumer hardware, like a MacBook Pro?