Skip to content

Mac Studio with 512GB RAM goes on sale for $22,299

Apple now sells the M5 Ultra Mac Studio with 512GB of unified memory, a $15,500 upgrade that lifts the price to $22,299. What you get and who it is for.

· · 3 min read

Share
Front of a silver Mac Studio desktop on a white background (earlier 2022 model, illustrative photo)
Image: Yasu / Wikimedia Commons, CC BY-SA 3.0
On this page

Apple has switched on the most expensive option in its desktop line-up. The Mac Studio with M5 Ultra can now be configured with 512GB of unified memory on Apple's online store, the option Apple said would arrive in late October when it launched the M5 Max and M5 Ultra models in August.

What it costs

The 512GB option is tied to a single chip: the M5 Ultra with a 36-core CPU, 80-core GPU and 32-core Neural Engine. According to MacRumors and 9to5Mac, that configuration starts at $6,799 with 96GB of memory, and the jump to 512GB adds $15,500. That puts the cheapest 512GB Mac Studio at $22,299 with a 1TB SSD. Max out storage at 16TB and the price reaches $29,799.

Macworld points out an odd quirk: the same machine with 256GB costs $10,799, so two 256GB Mac Studios together cost about $700 less than one 512GB unit, though of course that is two separate computers.

According to 9to5Mac, 512GB orders are quoted for delivery in late November to early December.

The specs that matter

Mac Studio M5 MaxMac Studio M5 Ultra
Base memory36GB96GB
Maximum memory128GB512GB (36-core CPU / 80-core GPU chip only)
Memory bandwidthLower than Ultra1.2TB/s (Apple)
Maximum storage8TB16TB
Configuration limits from Apple's Mac Studio tech specs page.

On Apple silicon the CPU, GPU and Neural Engine share one pool of unified memory. That is the real reason this option exists. For local AI, the size of the model you can load is limited by memory, not just compute, and 512GB is far more than any consumer graphics card offers. Our explainer on how much VRAM you need for local AI explains why memory is the bottleneck, and why model size and quantisation decide what fits.

Who is it for?

Almost nobody needs this. For video editing, music, coding and even most local AI experiments, the 96GB or 256GB configurations offer far better value. The 512GB option is aimed at developers and researchers who want to run very large open-weight AI models on one quiet desktop instead of a multi-GPU server. Meta, for example, recently said its 816B-parameter Muse Spark Open model runs on two Mac Studios.

If you are buying a Mac for everyday work, see our guide on how much RAM you actually need. Unified memory cannot be upgraded later, so buy what you will need for the next few years, but not much more.

Share

More from Computers & AI

View all →

Read next