Back to latest

Apple’s Autumn Surprise: Next-Gen M6 and M5 Ultra Chips Unleashed Alongside Upgraded Mac mini and Mac Studio

Sources Apple World Today iDownloadBlog.com iLounge | All Things iPod, iPhone, iPad and Beyond MacRumors MacSparky MacStories Macworld Mac Gems News and Updates - Apple …

Sources

Highlights

In a stunning late-August hardware surprise, Apple has unveiled its next-generation silicon family, debuting the industry’s first 2-nanometer M6 chip and the powerhouse M5 Ultra. These ground-breaking chips arrive inside completely refreshed Mac mini and Mac Studio models, bringing massive performance gains for on-device AI tasks but carrying higher price tags. The announcements coincide with a transitional moment for the company, coming just days before Tim Cook officially steps down as CEO after a historic 15-year run.

Top Stories

  • Apple Announces New M6 and M5 Pro Mac minis, Along with M5 Max and M5 Ultra Mac Studios: Apple has blindsided the industry with a rare August hardware release, updating both the Mac mini and Mac Studio lineups. The Mac mini is now configurable with either the new M6 or the M5 Pro, while the Mac Studio can be configured with the M5 Max or the behemoth M5 Ultra. Both lineups deliver incredible performance boosts but come with price hikes—$100 more for the mini and $200 more for the base M5 Ultra Studio. (Source)
  • Apple Reveals M6 as First-Ever 2nm Chip: Apple has officially taken the lead in chip fabrication by unveiling the M6, the industry’s first processor built on a cutting-edge 2-nanometer process. The chip features a redesigned 12-core CPU with two high-powered “super cores” and a 12-core GPU with Neural Accelerators built into each core. It also introduces a Dual 16-core Neural Engine that delivers up to twice the peak compute performance of prior generations to accelerate on-device AI model execution. (Source)
  • Apple Debuts M5 Ultra as Most Powerful Chip Ever: The M5 Ultra represents a massive leap in desktop computing, fusing two dual-die M5 Max chips with next-generation UltraFusion technology to form Apple’s first quad-die architecture. The chip scales up to an incredible 36-core CPU, an 80-core GPU with Neural Accelerators, and up to 1.2TB/s of unified memory bandwidth. When configured with up to 512GB of RAM in late October, the M5 Ultra allows professional workflows to run massive open-weight AI models completely on-device. (Source)
  • Apple Maps Ads Officially Roll Out in the US and Canada: Following testing throughout the summer, Apple has officially launched ads inside its native Maps app for users in the US and Canada. The sponsored spots, clearly labeled with a blue “Ad” badge, appear in search suggestions under Suggested Places and at the top of relevant search results. While Apple highlights strict, privacy-first user protections that decouple search and location interactions from Apple Accounts, users have no way to turn them off. (Source)
  • Dynamic Siri AI Waitlist Estimates Prepped for iOS 27 Launch: Apple is preparing to manage the extreme cloud server load associated with its upcoming Apple Intelligence features by integrating a waitlist directly into the release versions of iOS 27 and macOS 27. New code discovered in developer betas indicates that Apple will push personalized wait time estimates directly from its servers. Users will receive a push notification once access is granted to the advanced Siri AI features, which could take anywhere from one to two weeks based on beta feedback. (Source)

Articles Worth Reading

It’s Tim Cook’s last week on the job. Let’s review his career (Source) With Tim Cook scheduled to step down as Apple CEO on September 1, this retrospective takes an unflinching look at his historic 15-year tenure. On one hand, Cook is celebrated as an operational genius who transformed Apple into a financial juggernaut, generating monumental returns for long-term investors. On the other hand, the piece critiques how Apple’s relationship with developers and customers has grown increasingly transactional under his watch, citing aggressive subscription pushes and the integration of ads into once-pristine services like Maps. It’s a highly engaging and critical must-read for anyone reflecting on this epochal transition in Apple’s leadership.

The Potential of M6 and M5 Ultra for Local AI on macOS (Source) This detailed analysis breaks down the massive implications of Apple’s newly announced 2nm M6 and quad-die M5 Ultra chips for running advanced artificial intelligence models locally. The author, who manages their review writing workflows with local agents via Apple’s MLX framework, highlights the massive impact of Neural Accelerators on each GPU core. Under linear scaling assumptions, a Mixture-of-Experts model that runs at 17 tokens/second on an M4 Mac mini could soar past 60 tokens/second on the M6. Furthermore, the M5 Ultra’s support for up to 512GB of RAM and 1.2TB/s of bandwidth makes it an absolute beast for running frontier open-weight models entirely on-device, representing a massive shift for the industry.

Apple @ Work: Parallels Desktop 27 brings OpenGL 4.3 and AI acceleration to Apple Silicon (Source) Parallels Desktop 27 introduces game-changing virtualization enhancements that will bridge critical gaps for enterprise and developer workflows on Mac. The standout feature is a new Metal-based graphics driver offering hardware-accelerated OpenGL 4.3 support, allowing demanding Windows applications like Blender and ArcGIS Pro to run natively on Apple Silicon. Crucially for developers, the software adds support for Scalable Matrix Extension (SME2) instructions on M4 and newer Macs, resulting in up to 7x faster local AI operations. This article is highly recommended for IT administrators and developers looking to optimize cross-platform development environments.


📊 Nudge: Since there are some dramatic hardware price trends and comparison benchmarks in today’s news, I could generate a chart comparing the new M6 and M5-series Mac pricing against their predecessors to help you visualize the cost of upgrading.

Search MacWorks

Enter at least two characters.