On August 25, 2026, Apple Newsroom put M6 in a new Mac mini and, in a second note, put M5 Ultra in a new Mac Studio. Mini M6 starts at $899 US. Mini M5 Pro starts at $1,699 US. Studio M5 Max starts at $2,499 US. Studio M5 Ultra starts at $5,499 US. Pre-order is open. Both chassis start shipping September 22. This desk is not inventing CAD. Live Canadian list prices are on apple.com/ca.
What Apple actually announced
The Mac mini lineup is now M6 and M5 Pro. The Mac Studio lineup is now M5 Max and M5 Ultra. M6 is Apple's first 2-nanometer chip in this chassis: a 12-core CPU, a 12-core GPU with Neural Accelerators for the first time on mini, Dual 16-core Neural Engine, 16GB standard unified memory configurable up to 32GB, and up to 170GB/s bandwidth. Apple lists up to 4x faster AI versus the M4 mini, 2x storage and graphics versus M4, and 40 percent faster CPU versus M4. LLM prompt processing is listed at up to 13.5x versus the M1 mini and 4.8x versus M4.
Mini connectivity is Wi-Fi 7, Bluetooth 6, and 2.5Gb Ethernet with a 10Gb option. Front: two USB-C ports and a headphone jack. Back: three Thunderbolt 4 ports on M6, three Thunderbolt 5 ports on M5 Pro, HDMI, and Ethernet. M5 Pro goes up to an 18-core CPU, a 20-core GPU, 64GB unified memory, and 307GB/s bandwidth. Thunderbolt 5 can cluster minis for large on-device models.
"Mac mini has always been our most versatile Mac. Whether it's being used as a home computer, powering a professional studio, or as an always-on agentic device, it's the little Mac that can do it all."Johny Srouji, Apple's chief hardware officer, Apple Newsroom Mac mini, August 25, 2026
Studio with M5 Max is an 18-core CPU (6 super plus 12 performance), up to 40-core GPU, up to 128GB unified memory, up to 614GB/s bandwidth, and up to 5 displays. Studio with M5 Ultra is the first quad-die M-series: up to 36-core CPU, up to 80-core GPU, up to 512GB unified memory, 1.2TB/s bandwidth, a 32-core Neural Engine, and up to 8 displays. Apple says a cluster of four Studios is up to 3x faster at AI inference than one. Storage is PCIe Gen 6 SSD, listed up to 2x faster. Thunderbolt 5 is up to 120Gb/s. The N1 chip brings Wi-Fi 7 and Bluetooth 6 to Studio for the first time.
What it costs
These are Apple's published US list prices from the August 25 newsroom notes. Do not quote them as CAD. Price the configuration on apple.com/ca the day you buy.
| Chip | Start US | Education US | Who it is for |
|---|---|---|---|
| Mac mini M6 | $899 US | $799 US | Local agents on a small desk |
| Mac mini M5 Pro | $1,699 US | $1,599 US | Pro work; cluster minis for larger models |
| Mac Studio M5 Max | $2,499 US | $2,299 US | When the model needs the memory (up to 128GB) |
| Mac Studio M5 Ultra | $5,499 US | $5,099 US | Frontier on-device models (up to 512GB) |
Pre-order opened August 25. Both chassis ship September 22. The Studio 512GB unified memory option is late October. Apple's US Upgrade lease lists Studio M5 Max from $48.99/mo and Ultra from $110.10/mo over 36 months. That is a US lease, not an Alberta quote.
Apple.ca lists these base configurations. This desk is not attaching invented CAD to them: Mini M6 12-core CPU / 12-core GPU, 16GB, 256GB; Mini M5 Pro 15-core CPU / 16-core GPU, 24GB, 512GB; Studio M5 Max 18-core CPU / 32-core GPU, 36GB, 512GB; Studio M5 Ultra 30-core CPU / 64-core GPU, 96GB, 1TB.
What the chips say about on-device work
Apple says M5 Ultra lets users run huge large language models with hundreds of billions of parameters entirely on the device. That is an Apple compute claim, not a promise that a shop's client files stay private. The Mac mini product page also mentions clustering Mac mini devices with exo and LM Studio Bionic for multiple AI agents. For a Sherwood Park or Edmonton shop that already keeps work on a desk, that is the useful sentence: a mini can run agents without sending a tender to a nameless cloud. The Studio is the box for bigger local models. See the Mac mini AI agent path if the next step is a local stack, not a new subscription.
Apple Intelligence features are currently available for testing through the Apple Beta Software Program and will be available with macOS 27 this fall. Siri AI is coming in English later this year as a beta with macOS 27. This briefing does not treat either as shipping, complete, or equivalent to a private operator stack.
What Alberta operators should do
If the job is a handful of agents on a private desk in Sherwood Park or Edmonton, start with the mini and the published M6 numbers. If the job is a large on-device model, look at Studio when the model needs the memory. Pre-order is open against a September 22 start. The 512GB Studio SKU is late October. Do not convert US list prices into CAD in a quote. Price the configuration on apple.com/ca the day you buy.
Keep the files that would hurt if they leaked on a private AI path. A faster chip does not change who can read a prompt that left the building.
The guardrail
Chip speed is not a data-residency policy. On-device work stays on-device only if the workflow stays on-device. Clustering, cloud sync, and a helpful assistant that phones home are separate choices. For AI consulting work, treat M6 and M5 Ultra as published silicon, not as a compliance stamp. Keep client files on a private path when a leak would hurt.