Apple today introduced the M6 and M5 Ultra, marking the company’s first foray into 2-nanometer chip design and the debut of quad-die architecture in its M-series lineup. The M6 arrives in a new Mac mini launching September 22, while the M5 Ultra powers an upgraded Mac Studio available for pre-order today with general availability starting September 22; configurations with up to 512GB of memory ship in late October.
The M6 represents Apple’s most granular CPU design yet, pairing two super cores with four performance cores and six efficiency cores for the first time across its desktop chips. It delivers up to 1.2x faster multithreaded performance than the M5 and up to 2.4x faster than the M1. GPU compute for AI workloads jumped 30 percent versus M5 and more than 8x versus M1. The chip supports up to 32GB of unified memory with 170GB/s of bandwidth, a 10 percent increase over M5 and 2.5x over M1.
The M5 Ultra is Apple’s largest chip to date, formed by connecting two dual-die M5 Max processors using next-generation UltraFusion technology. This quad-die setup yields a 36-core CPU (12 super cores, 24 performance cores) and an 80-core GPU, with unified memory bandwidth reaching 1.2TB/s, 50 percent higher than the M3 Ultra. Single-threaded performance climbs 1.25x and multithreaded performance 1.3x over M3 Ultra. UltraFusion delivers over 4.4TB/s of inter-die bandwidth with 6x higher connection density than previous solutions, allowing the four dies to operate as a single processor.
Mac Mini M6 and Mac Studio M5 Ultra Performance Gains
The Mac mini with M6 achieves up to 4x faster AI performance, 2x faster graphics and storage, and 40 percent faster CPU speed compared to its predecessors. In LM Studio, the M6 delivers up to 13.5x faster LLM prompt processing than the original M1 Mac mini and 4.8x faster than the M4. Both new Mac mini models gain Wi-Fi 7 and Bluetooth 6, stepping up from Wi-Fi 6E and Bluetooth 5.3.
Mac Studio with M5 Ultra posts 4.3x faster AI performance, 2x faster storage, 1.8x faster graphics, and 1.3x faster CPU speed. The machine can run models locally using the full 512GB unified memory pool. Its next-generation SSD architecture uses PCIe Gen 6 for twice the speed of the previous version, paired with Thunderbolt 5 ports delivering up to 120Gb/s of bandwidth, Wi-Fi 7, and Bluetooth 6. The M5 Ultra includes a 32-core Neural Engine and enhanced Media Engine capable of simultaneously playing up to 33 streams of 8K ProRes 422 at 30 fps.
What The M6’s 2-Nanometer Transition Means
The M6 is Apple’s first processor built on 2-nanometer manufacturing, a generational leap from the larger process nodes of previous chips. The smaller transistor density allows Apple to pack more processing capability into the same thermal envelope, directly enabling the CPU core configuration and GPU compute gains cited above. Sri Santhanam, Apple’s vice president of Silicon Engineering Group, called it “the next giant leap in performance and AI compute for Apple Silicon.”
The inclusion of all three CPU core types in a single chip is new for M-series desktop processors. This heterogeneous design lets the chip assign super cores to demanding single-threaded tasks like code compilation and image editing, performance cores to multithreaded workloads, and efficiency cores to background processes. Johny Srouji, Apple’s chief hardware officer, stated the M6 advances “every compute block, delivering gains across every dimension of performance.”
Developer Tools and AI Workload Implications
Apple emphasized that its developer frameworks, Core AI, Core ML, Metal, and Xcode, now directly tap into the M6 and M5 Ultra’s advanced capabilities. Developers can use the M6’s Dual 16-core Neural Engine and the M5 Ultra’s 32-core Neural Engine, combined with the massive 512GB memory pool and 1.2TB/s bandwidth on the high-end machine, to deploy sophisticated AI compute entirely on-device. This capability, paired with the machines’ positioning as “always-on agentic devices,” signals Apple’s strategic pivot toward local large language model inference as a primary use case rather than a secondary feature.
For content creation, Mac Studio with M5 Ultra handles complex timelines with multiple streams of high-resolution video and real-time effects in Adobe Premiere Pro. The doubled video encode and decode blocks compared to M5 Max unlock professional video workflows at scale.
Sustainability and Manufacturing
The new Mac Studio is assembled using 35 percent recycled content overall, including 100 percent recycled aluminum in the enclosure and 100 percent recycled rare earth elements in all magnets. Manufacturing employs 40 percent renewable energy across the supply chain.
The M6 and M5 Ultra represent a clear strategic commitment to owning high-end on-device AI compute through custom silicon. The M5 Ultra’s quad-die architecture and half-terabyte memory support would have seemed impossible in a desktop machine two years ago; the M6’s 2-nanometer process and threefold core-type design show Apple is willing to commit manufacturing complexity to deliver both consumer-friendly and pro-grade chips from the same architecture family. Availability begins September 22 for both machines, with Mac Studio’s maximum-memory configurations following in late October.
