> I’m running the same model on a 48GB MBP with a q4 quant and it’s pretty decent.

Context size?