Apple launches M6 MacBook Pro and iMac with up to 13x faster on-device LLM performance

Apple has announced a new 14-inch MacBook Pro and a refreshed iMac, both built around its latest M6 chip, with the company claiming its largest gains in on-device AI performance since the transition to its own silicon. According to TechCrunch, the announcement landed on Tuesday, one day ahead of Apple's "Pro and Tell" event scheduled for Wednesday, and follows an iPad mini refresh earlier in the week.

What the M6 delivers

According to Apple's press release, the M6 pairs a 12-core CPU — two cores more than the M5 — with a 12-core GPU that also gains two cores, a third-generation ray-tracing engine and Neural Accelerators built into every GPU core. A dual 16-core Neural Engine sits alongside them, which Apple says is up to 1.9x faster than the one in the M5. Apple also claims the CPU includes the fastest CPU core in the world.

Relative to the M5, Apple quotes up to 30% faster graphics and up to 25% faster AI workloads such as LLM prompt processing. TechCrunch reports that unified memory starts at 16GB and is configurable to 32GB, with memory bandwidth rising to 170GB/s from 153GB/s on the M5.

On-device AI is the headline

The gap versus older machines is where the marketing leans. Apple says the M6 MacBook Pro reaches a first token up to 13x faster when processing LLM prompts than the 13-inch MacBook Pro with M1, and up to 1.3x faster than the 14-inch M5 model. Other published figures include up to 11.7x faster ray-traced renders in Blender and up to 2.5x faster Xcode project builds versus the M1 machine.

The iMac follows the same pattern. Apple's product page claims it is up to 5x faster than the M4 iMac for on-device AI, up to 12x faster than the M1 model and up to 18x faster than the most popular Intel-based iMac.

Apple is explicitly positioning the Mac as a platform for local AI work. The iMac page highlights AI apps including Claude, LM Studio, Gemini and Topaz Labs, while the press release points to Apple Intelligence, Siri AI and locally run LLMs as core use cases, with privacy preserved by keeping processing on the device.

Hardware, pricing and availability

The MacBook Pro starts at $1,999 in space black or silver. It supports Wi-Fi 7 and Bluetooth 6, and Apple says its up-to-24-hour battery life is the longest ever in a Mac. TechCrunch lists the port selection as HDMI, MagSafe 3, three Thunderbolt 4 ports, an SDXC card slot and a headphone jack, along with fast charging that reaches 50% in 30 minutes using a compatible high-wattage adapter.

The iMac starts at $1,499 and comes in seven colors. It features a 24-inch 4.5K Retina display, up to four Thunderbolt 4 ports, support for two external 6K displays, optional 2.5Gb Ethernet and an optional nano-texture glass finish. Both computers ship with macOS 27.

Preorders open October 28 and both machines go on sale November 6. Apple's website also shows a new Mac mini offered with either the M6 or an M5 Pro chip, though that model was not covered in Tuesday's announcement.

Why it matters

Local AI inference has quietly become the specification that sells computers, and Apple is now benchmarking its chips in terms of LLM token generation rather than only graphics and compile times. The design choices behind the M6 — Neural Accelerators inside the GPU, a dual Neural Engine and higher memory bandwidth — reflect hardware increasingly shaped around generative workloads.

The framing is aimed squarely at holdouts: gains over the M5 are a modest 1.3x in most real-world tasks, but the comparison to M1 and Intel machines is dramatic, and those users make up a large share of the installed base. For developers and researchers running models locally, the 32GB memory ceiling remains the practical constraint on what can be loaded, whatever the Neural Engine can do. Apple's bet is that privacy plus battery-efficient inference keeps on-device AI, and the Mac, relevant as cloud-based alternatives multiply.