For the last decade the Mac Mini has been on a two year October release cycle.
I’m expecting it to go to M6 /. M6 Pro in October.
I’m expecting it to go to M6 /. M6 Pro in October.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
When is the Next Apple Event?
- Thread starter MacRumors
- Start date
- Sort by reaction score
Back in November we got rumors that the new MacBook Pro were referenced for MacOS 26.3, and in fact they got them out with MacOS 26.3
The MacStudio was referenced for 26.4, so here we are
The MacStudio was referenced for 26.4, so here we are
Night_Sailor
macrumors newbie
One interesting item. I noticed the M3 Ultra Studio is no longer available with 512 gb of unified memory. It a separate discrete component. It seems like they are hoarding memory for the M5 Ultra Studio. Good!
That m5 UltraStudio would make it a stout AI machine capable of locally loading very large language models without the huge power needs of a PC with a power hog Nvidia graphics card. I am already planning my own LLM for a nitch market AI service.
If 1 TB of unified memory is available I’ll definitely be buying one. It will be impossible to resist. 512 will make me upgrade.
There are benefits to a larger and pricier SSD also, specifically faster loading of LLM’s and other apps and other datasets. Nothing external will be as fast. I’ve been rebuilding my home network entirely focused on VLANs to segment and secure it—in part for improved security and primarily to protect it from AI agents. Would it need to be maxed out? Probably not, but then again why not? I’m guessing $11,000 for a maxed out Studio.
I considered a 96GB graphics card for my windows machine, but the cost will be about the same as a tricked out Studio system. Some people are runniing dual cards. Others are multiplexing multiple Mac Studios.
Dell has their GB10 and GB300. The later will require the purchase of a home nuclear reactor to supply power. LOL.
The GB10 or equivalent systems may be on the table for me, and while I love the idea of the GB300, the power requirements have me thinking about a solar array to supply it.
Which leads me back the the m5 Ultra Studio. It will be efficient enough to leave on and powerful enough to load the larger LLM’s. All the new tech demonstrated in the M5 MacBook Pro leads me to believe other option willl perform better but the Ultra will win on processing per Watt.
That m5 UltraStudio would make it a stout AI machine capable of locally loading very large language models without the huge power needs of a PC with a power hog Nvidia graphics card. I am already planning my own LLM for a nitch market AI service.
If 1 TB of unified memory is available I’ll definitely be buying one. It will be impossible to resist. 512 will make me upgrade.
There are benefits to a larger and pricier SSD also, specifically faster loading of LLM’s and other apps and other datasets. Nothing external will be as fast. I’ve been rebuilding my home network entirely focused on VLANs to segment and secure it—in part for improved security and primarily to protect it from AI agents. Would it need to be maxed out? Probably not, but then again why not? I’m guessing $11,000 for a maxed out Studio.
I considered a 96GB graphics card for my windows machine, but the cost will be about the same as a tricked out Studio system. Some people are runniing dual cards. Others are multiplexing multiple Mac Studios.
Dell has their GB10 and GB300. The later will require the purchase of a home nuclear reactor to supply power. LOL.
The GB10 or equivalent systems may be on the table for me, and while I love the idea of the GB300, the power requirements have me thinking about a solar array to supply it.
Which leads me back the the m5 Ultra Studio. It will be efficient enough to leave on and powerful enough to load the larger LLM’s. All the new tech demonstrated in the M5 MacBook Pro leads me to believe other option willl perform better but the Ultra will win on processing per Watt.
