Apple announced M6 and M5 Ultra desktop systems aimed partly at local AI workloads, including a Mac Studio configuration with up to 512 gigabytes of unified memory. New Mac mini and Mac Studio configurations emphasize memory capacity and bandwidth for running models without sending prompts to cloud services. The dated account begins with this confirmed point: The M6 is Apple’s first two-nanometer Mac chip.
The M6 configuration offers up to 32 gigabytes of memory. The M5 Ultra offers up to 36 CPU cores and 80 GPU cores. Those details establish the reported scale and the institutions directly involved, while keeping statements from officials, companies, hospitals or prosecutors attached to the people and records that supplied them.
The M5 Ultra configuration supports up to 512 gigabytes of unified memory and 1.2 terabytes per second of bandwidth. Apple set starting prices of 899 dollars for Mac mini and 2,499 dollars for Mac Studio. The combination describes the immediate operating picture at publication time; it does not convert a preliminary explanation, allegation, forecast or organizational announcement into a settled final outcome.
Preorders opened August 25 and most systems were scheduled to ship September 22. The highest-memory configuration was expected later in October. That background places the new event in its relevant chronology and explains why some figures or procedures differ from the simpler version conveyed by a headline.
Unified memory capacity affects which model weights and context can remain resident, but does not by itself establish inference speed. Performance claims had not yet been tested in independent production benchmarks. The unresolved boundary is part of the factual record because later investigation, technical testing, audited data and newly available source access can materially revise early accounts.
For this edition, the available evidence on Apple Repositions Mac Desktops for Local AI Inference comes from Ars Technica. The reporting uses those records for dated events and attributed statements, without treating an institution’s description of its own action as independent proof of every underlying claim.
The next observable developments are third-party model and power-efficiency benchmarks and software support for the largest local models. Until either produces a new attributable record, the known sequence remains The M6 is Apple’s first two-nanometer Mac chip. The M6 configuration offers up to 32 gigabytes of memory. The M5 Ultra offers up to 36 CPU cores and 80 GPU cores.
A second reading of the evidence also keeps three categories separate: the direct event (The M6 is Apple’s first two-nanometer Mac chip.), the measured or reported consequence (The M5 Ultra configuration supports up to 512 gigabytes of unified memory and 1.2 terabytes per second of bandwidth.), and the remaining unknown (Performance claims had not yet been tested in independent production benchmarks.). This separation avoids implying a cause, motive or forecast that the checked material does not establish.