Apple Inc. debuted its first 2-nanometer chip, M6, in the new Mac mini and introduced the updated Mac Studio featuring M5 Max and M5 Ultra silicon. The launch delivers major gains in on-device AI performance, with the M5 Ultra offering up to 4.3x faster AI compute than its predecessor.

The M6 chip is built on cutting-edge 2nm process technology, packing greater transistor density into a smaller die. It features a 12-core CPU complex with two super cores, four performance cores, and six efficiency cores. This architecture delivers up to 1.2x faster multithreaded performance compared to M5 and up to 2.4x faster than M1. The chip also includes a Dual 16-core Neural Engine, providing up to 2x the peak compute over previous generations, and supports up to 32GB of unified memory with 170GB/s bandwidth. The GPU features Neural Accelerators in each core, delivering a nearly 30% increase in peak GPU compute for AI compared to M5.

Mac Studio: M5 Max and M5 Ultra

The new Mac Studio features M5 Max and the all-new M5 Ultra, designed for demanding pro workflows and local AI inference. The M5 Max includes an 18-core CPU and an up-to-40-core GPU with Neural Accelerators in each core, supporting up to 128GB of unified memory. It delivers up to 3.9x faster AI performance than the prior generation.

The M5 Ultra uses next-generation UltraFusion technology to form a quad-die architecture, connecting two dual-die M5 Max chips with over 4.4TB/s inter-die bandwidth. It features an up-to-36-core CPU (12 super cores, 24 performance cores) and an up-to-80-core GPU, supporting up to 512GB of unified memory and 1.2TB/s of bandwidth. This configuration enables users to run large language models (LLMs) with hundreds of billions of parameters entirely on device, achieving up to 4.3x the peak AI compute performance of M3 Ultra. The GPU also incorporates hardware-accelerated mesh shading and third-generation ray tracing.

Application benchmarks

Apple provided specific benchmark comparisons for both chips across key applications.

M6 benchmarks:

• LLM prompt processing in LM Studio: up to 13.5x faster than Mac mini with M1; up to 4.8x faster than M4
• Microsoft Excel spreadsheet calculations: up to 2.3x faster than M1; up to 1.5x faster than M4
• Gaming in Cyberpunk 2077 with ray tracing: up to 2x faster than M4

Mac Studio with M5 Max benchmarks:

• LLM prompt processing in LM Studio: up to 10.7x faster than Mac Studio with M1 Max; up to 3.9x faster than M4 Max
• Text-to-image performance: up to 7.4x faster than M1 Max; up to 3.5x faster than M4 Max
• Magic Mask in DaVinci Resolve Studio: up to 5.3x faster than M1 Max; up to 3x faster than M4 Max

Mac Studio with M5 Ultra benchmarks:

• CopyCat ML training in Foundry Nuke: up to 15.4x faster than M1 Ultra; up to 3.3x faster than M3 Ultra
• LLM prompt processing in LM Studio: up to 9.8x faster than M1 Ultra; up to 4x faster than M3 Ultra
• Text-to-image performance: up to 8.2x faster than M1 Ultra; up to 4.3x faster than M3 Ultra

Connectivity and software

Both Mac mini models support Wi-Fi 7 and Bluetooth 6. The M6 model includes three Thunderbolt 4 ports, while the M5 Pro variant features three Thunderbolt 5 ports. The new Mac Studio also introduces Wi-Fi 7 and Bluetooth 6 for the first time, along with Thunderbolt 5 ports delivering up to 120Gb/s bandwidth. This connectivity allows multiple Mac Studio systems to be clustered using RDMA, creating a shared memory pool that delivers up to 3x faster AI inference compared to a single system.

The devices launch with macOS 27 Golden Gate, which includes Siri AI and next-generation Apple Intelligence features. Developers can leverage Core AI, Core ML, Metal, and Xcode to build and deploy AI models directly on Apple silicon, utilizing the Neural Accelerators and massive unified memory pools.

Pricing and availability

• Pre-orders begin August 25 on apple.com/store and the Apple Store app in 30 countries
• Availability begins September 22 at Apple Store locations and authorised resellers
• Mac Studio with 512GB of unified memory is coming in late October
• The new Mac mini is made with 50% recycled material overall, while the Mac Studio uses 35% recycled content overall, including 100% recycled aluminium in the enclosure

Apple Inc. is reducing headcount across its Siri, Vision Pro immersive video, and gaming divisions. The company announced the workforce adjustments on August 21, 2026, targeting specific teams within its broader technology portfolio. This move signals a continued refinement of its product strategy as it navigates evolving market demands for spatial computing and artificial intelligence features.

The layoffs affect employees working on core functionalities of Apple's virtual assistant and its mixed-reality headset ecosystem. While no specific number of positions was disclosed, the cuts span multiple high-profile units including the development of immersive video content and gaming experiences for the Vision Pro platform.

Strategic Focus

The decision to trim staff in these areas comes as Apple seeks to optimize resources across its innovation pipeline. The Siri team, responsible for the company's voice assistant, faces restructuring alongside the groups developing software and content for the Vision Pro headset.

This realignment suggests a prioritization of certain technological initiatives over others. By reducing headcount in gaming and immersive video, Apple may be consolidating efforts around more immediate revenue drivers or refining its approach to spatial computing applications.

What the Numbers Show

While specific financial figures were not disclosed in the announcement, the targeted nature of the cuts indicates a strategic shift rather than a broad-based cost reduction. The focus on Siri and Vision Pro-related teams highlights areas where Apple is actively recalibrating its investment strategy.