[pram]

FWIW the M5 appears to be an actual large leap for LLM inference with the new GPU and Neural Accelerator. So id wait for the Pro/Max before jumping on M3 Ultra.

[mitjam]

Thanks, that helps me keep things in perspective.