I read through the GS forum and was a bit disheartened. However, from that forum it looks like it's something fixable through an update.
Posted by azrulsaleh on gearslutz:
"I can confirm that the latency is not due to hardware issue but related to CoreAudio. Tested at 128 samples of buffer size and I get the following roundtrip results with different i/o settings:
CoreAudio: 19.5ms
Soundflower: 5.8 ms
Boom: 5.8 ms"
Hopefully I'm not breaking any forum rules by quoting another forum.
Not surprising this one slipped through whatever QA there is in this area of development though, since most people worried about this sort of thing probably use an interface and headphones. BUT, there shouldn't be a downgrade in a new iteration of a machine (opening a can of worms here with other specs, I know).
As a disclaimer, I am not a professional and have only used GarageBand. But I hope to delve deeper into the home recording scene once I get my new rMBP (I've been computerless since my 2008 white macbook took a poo,) so I hope I will see some of you around!