Here is what Mark Gurman was forecasting as of last month (July)
www.macrumors.com
He also missed the Mac Studio introduction when he predicted the Mini was incoming in his Monday report. The July report had very few specifics and shouldn’t count toward a true leak. (And people **** on him, not because he doesn’t have good sources or usually hits the mark on his predictions, it’s in how absolutely petty and childish he gets when facing an ounce of criticism over the (admittedly few) calls he misses. It’s a common trait amongst those lacking in humility.)
Been running the same calculation and I'm leaning towards the Mac Studio M5 Max - its not crazy more expensive that the maxxed Mini, but the M5 Max has crazy wide memory bandwidth compared to the M5 Pro.
Great argument. Problem is; people keep saying this over and over again under Gurman's reports and their subsequent confirmation. If every time he reports on something you have to say "broken clock is right twice a day", maybe your clock isn't broken.
That's not entirely true. First of all, they share much wider decode + they have their own private 1MB cache. That's why M4, M3 are not the same in that regard. It's not simple renaming.
The M7 is likely going to hit the market much sooner than that. Probably Spring 2027.
This is probably one of the biggest leaps since the M1 and this speed is only going to accelerate to meet demand.
The majority of new Mac Mini users are pushing the limits of the machine and these will accelerate that further. The Mac Mini has become the go-to machine for on-device LLMs and the M4 Mac Mini, while still OK, was struggling already, especially with larger models.
Most of these people don't work with large data sets and they think (as one commentor wrote) it is just so people can =sum two cells.
M4 p-cores issue width: 10
In the same way you find my lack of trust in Gurman strange, I find your defense of him odd. Are you his mommy?
I’ll give half of a Gurman was wrong. It looks like he’s right about there not being an M6 Pro, Max or Ultra, but not for the reason he said (which is where he’s most often “wrong”). Looking at the specs of the M6, most of the benefits seem to be around adding features that the M5 Pro and up chips got, such as neural accelerators in the GPU cores and these new “performance” cores.
Neural accelerators was the new thing of Apple 10 GPU used in both A19 and M5. A20/M6 are still using the same GPU. A21/M7 will introduce Apple 11. The only “new” thing is the performance cores, but they was still part of the M5 family.
The naming almost makes sense with all three in one chip, but it’s still going to drive years of incorrect comparisons like this for years to come. They really needed to pick a different name rather than reuse it from a better tier they already used for years.
