FULL STORY

Apple Launches M5 Ultra and M6 Chips for On-Device AI

From Gurman's leak to the official August 25 launch, Apple unveiled M5 Ultra and 2nm M6 chips plus new Mac minis focused on local AI inference.

2026-08-25 ~ 2026-08-26 · 4 episodes · 18 posts

Episode 1 · Apple Reportedly Set to Launch New Mac Mini for On-Device AI (2026-08-25, 2 posts)

Apple is expected to unveil a new Mac mini within days, ahead of its September iPhone event, targeting surging demand for on-device AI. Analysts note that sufficient memory for local models could drive prices higher.

Episode 2 · Apple Unveils M6 and M5 Ultra Chips, Bets Big on Local AI (2026-08-25, 6 posts)

On August 25, Apple officially unveiled the M5 Ultra and M6 chips alongside new products, touting major upgrades in on-device AI and graphics performance. Apple calls the M5 Ultra "the most powerful chip ever," built on a quad-chip architecture that fuses two M5 Max dies via the new Ultra Fusion design, with inter-chip bandwidth reaching 4.4 (unit incomplete in the original post), up to 512GB of unified memory, and 1.2TB/s memory bandwidth—a 50% boost over the previous-generation M3 Ultra. The new Mac Studio with M5 Ultra starts at $5,499 (with 1TB of storage) and can run large language models with hundreds of billions of parameters entirely on-device; a lower-end configuration with M5 Max is also available. Also announced, the M6 is Apple's first 2nm chip, featuring a 12-core CPU, 12-core GPU, and dual 16-core neural engines, debuting in the Mac mini starting at $899.

Confirmed

  • The M5 Ultra uses a quad-chip Ultra Fusion architecture combining two M5 Max dies, with up to 512GB unified memory and 1.2TB/s bandwidth, a 50% increase over the M3 Ultra
  • The new Mac Studio comes in M5 Max and M5 Ultra versions; the M5 Ultra model starts at $5,499 (1TB storage)
  • The M6 is Apple's first 2nm chip with a 12-core CPU, 12-core GPU, and dual 16-core neural engines, powering the Mac mini starting at $899
  • Apple positions the M5 Ultra as a flagship chip optimized for 3D rendering and running frontier AI models

Why it matters

  • 512GB of unified memory makes fully local inference of LLMs with hundreds of billions of parameters possible, significantly reducing reliance on cloud computing and marking a key step in the on-device AI race
  • The M6's 2nm process signals Apple's continued lead in advanced manufacturing, paving the way for future performance and efficiency gains across the Mac lineup

Episode 3 · Apple Launches M6 Mac Mini as Local AI Inference Machine (2026-08-25, 8 posts)

On August 25, Apple officially released the new Mac Mini with M6 and M5 Pro chips, with the M6 model starting at $899, alongside the Mac Studio featuring M5 Max and M5 Ultra chips. The takeaway: Apple has clearly positioned the Mac Mini M6 as consumer hardware for running local AI agents and on-device inference — a major step in bringing Apple Silicon to edge AI.

Confirmed

  • The M6 model delivers up to 4x higher AI performance (vs. the previous M4 Mac mini), doubles graphics and storage speeds, boosts CPU performance by 40%, and supports Wi-Fi 7 and Bluetooth 6
  • Per @xiaohu, the M6's 12-core GPU adds a neural accelerator to every core for the first time, with peak AI compute nearly 30% higher than M5 and over 8x higher than M1, mainly to speed up LLM prompt processing
  • Per @flavioAd, Apple officially markets the Mac Mini M6 as a computer for local AI agents and on-device inference
  • Per @umeshai, the new model is more compact while handling both everyday productivity and AI workloads
  • Per @bytebot, the M6 and M5 Pro Mac Mini models carry higher starting prices; Apple's marketing focus has shifted from the Neural Engine to outright AI leadership, a change that began with M5

Why it matters

  • Consumer hardware is being formally cast as local AI infrastructure, signaling that LLM inference is accelerating its move to the edge
  • Apple's AI-first pricing and marketing shift could intensify the AI race across the entire desktop hardware market

Episode 4 · Rumor: Apple M6 Gets Dual Neural Engines for Big AI Boost (2026-08-25, 2 posts)

Leaks suggest Apple's upcoming M6 chip will feature dual 16-core neural engines and fp8 GPU matrix support, delivering roughly 4.8x faster LLM prompt processing than M4 and enabling local 70B-parameter models.