Pre-orders opened the same day, with availability starting September 22, 2026 (the highest-memory Mac Studio configurations arrive later, in late October). The machines ship with macOS 27 (Golden Gate) and support the next generation of Apple Intelligence, including Siri AI.
These updates come amid strong demand for Macs as local AI platforms (users have been daisy-chaining prior models to run larger models privately and avoid cloud token costs). Designs remain largely unchanged—the compact Mac mini footprint and Mac Studio form factor are the same—while internal performance, AI accelerators, memory options, storage speeds, and some connectivity improve. Prices have risen, reflecting higher component costs (especially memory) driven by AI industry demand.
New Mac mini (M6 and M5 Pro)
The base model introduces Apple’s first 2-nanometer process chip, the M6. It features a 12-core CPU (2 super cores + 4 performance cores + 6 efficiency cores), a 12-core GPU with Neural Accelerators in every core, a dual 16-core Neural Engine (32 cores total, doubling prior peak AI compute in some comparisons), and up to 170 GB/s unified memory bandwidth. Apple claims roughly 40% faster CPU performance vs. the prior M4 Mac mini, up to 2x graphics and storage speeds, and up to 4x faster AI performance. Memory starts at 16 GB (configurable up to 32 GB on M6 configurations).
A higher-end option uses the M5 Pro (already seen in recent MacBook Pros): up to an 18-core CPU and 20-core GPU, with higher memory support (up to 64 GB). This targets more demanding pro tasks such as visual effects. Storage options scale higher, and Ethernet starts at 2.5 Gb (upgradable to 10 Gb). The M6 model retains Thunderbolt 4, while the M5 Pro gains Thunderbolt 5. Wi-Fi 7 and Bluetooth 6 are included.
Pricing: M6 starts at $899 (16 GB / 256 GB); M5 Pro starts at $1,699. These are $100 higher than the most recent prior starting prices. Configurations can climb substantially with more memory and storage.
The Mac mini is positioned as an always-on, energy-efficient machine for agentic AI, local models, coding, creative work, and everyday productivity in a tiny chassis.
New Mac Studio (M5 Max and M5 Ultra)
This is marketed as the ultimate desktop for on-device AI and extreme pro workflows. The M5 Max configuration offers an 18-core CPU, up to a 40-core GPU with Neural Accelerators in each core, and up to 128 GB unified memory. It delivers major gains in LLM prompt processing, text-to-image generation, video tools (e.g., Magic Mask in DaVinci Resolve), and other pro apps versus prior Max chips.
The flagship M5 Ultra is Apple’s first quad-die M-series chip (via next-generation UltraFusion interconnecting two dual-die M5 Max dies). It scales to a 36-core CPU (12 super + 24 performance), up to an 80-core GPU (first Ultra with Neural Accelerators), a 32-core Neural Engine, up to 512 GB unified memory, and 1.2 TB/s memory bandwidth (50% higher than M3 Ultra). Apple claims up to 4.3x peak AI compute vs. M3 Ultra (and much higher vs. earlier Ultras), along with strong gains in graphics, CPU multithreaded performance, and media engine capabilities (e.g., many concurrent 8K ProRes streams). Storage is up to twice as fast.
Notable features include official support for clustering multiple Mac Studios over Thunderbolt 5 with RDMA to pool memory—Apple says four systems can deliver up to 3x the AI inference throughput of one. New software includes the Core AI framework (alongside MLX) for building, running, and deploying models optimized for Apple silicon’s unified memory, CPU, GPU, and Neural Engine. Connectivity upgrades: Wi-Fi 7 and Bluetooth 6 (via Apple’s N1 chip) arrive for the first time on Mac Studio, alongside Thunderbolt 5, support for up to eight displays, and other pro I/O.
Pricing: M5 Max starts at $2,499; M5 Ultra starts at $5,499 (a $200 increase over the prior Ultra base in some comparisons). High-end configs (especially with 512 GB memory) are far more expensive.
Broader Context and Implications
- AI focus: Both machines emphasize private, on-device inference of large/open-weight models, agentic workflows, and avoiding cloud costs/privacy trade-offs. The Mac Studio’s memory scale and clustering make it particularly relevant for researchers, developers, and teams working with frontier models. The Mac mini suits always-on agents or lighter local AI.
- Performance claims are Apple’s (relative to prior Macs in specific apps/tools like LM Studio, DaVinci Resolve, Redshift, etc.); real-world results will vary with software optimization, model size, and configuration. Neural Accelerators in GPU cores and dual Neural Engines are key to the AI speedups.
- Price and availability pressures: Higher starting prices reflect memory/component cost inflation from AI data-center demand. Prior-generation Mac minis and Studios saw strong demand (and shortages) for local AI use; these refreshes may face similar constraints initially.
- Software ecosystem: macOS 27, Apple Intelligence/Siri AI, Core AI, MLX, Xcode, and third-party tools (LM Studio and others) are central. Unified memory remains a major architectural advantage for AI workloads on these systems.
- Who they’re for: Mac mini for compact, efficient everyday/AI-agent use and mid-tier pro work; Mac Studio for high-end creative (video, 3D/VFX, audio), scientific, software development, and serious local AI/ML. Both pair with Studio Display options.
For full official details, see Apple’s Newsroom pages on the Mac mini, Mac Studio, and the M6/M5 Ultra chips. Specs, exact performance multipliers, and configure-to-order options are available on apple.com. As with any new silicon launch, i
