iMac with M6 is Apple’s new 24-inch all-in-one, announced on Tuesday 27 October 2026, and Apple is selling it mainly on artificial intelligence. Its press release puts the claim in a heading, “A Huge Leap in Performance and AI Capabilities”, and backs it with one striking number: up to 5x faster on-device AI than the iMac with M4 it replaces.

The rest of the machine changes less. The 4.5K display, the thin design and the seven colours carry over. What is new sits inside: Apple’s first 2-nanometre chip, a Dual 16-core Neural Engine, Neural Accelerators in every GPU core, faster memory, Wi-Fi 7 and optional 2.5Gb Ethernet. Pre-orders open on 28 October, and the first machines arrive on Friday 6 November.

This guide explains what changed, what the 5x figure measures and what it does not, which local AI models the iMac with M6 can realistically run, and why the cheapest model now costs $200 more than it did. We finish with who should upgrade and what businesses should check first.

What Apple Announced on 27 October

Apple published the news as a press release, titled “Apple unveils new iMac with M6, delivering a major boost in performance and AI capabilities”. Engadget’s report the same morning carried the headline this article uses, and noted that the extra speed is “much more pronounced” for AI than for everyday work.

The short version

  • Chip: M6, with up to a 12-core CPU, up to a 12-core GPU with Neural Accelerators, and a Dual 16-core Neural Engine.
  • AI claim: up to 5x faster on-device AI than the M4 model, and up to 12x faster than the iMac with M1.
  • Memory: 16GB as standard, up to 32GB, with up to 170GB/s of bandwidth.
  • Connectivity: Wi-Fi 7 and Bluetooth 6, plus 2.5Gb Ethernet as an option.
  • Price: from $1,499 or £1,499 for the 10-core model and $1,699 or £1,699 for the 12-core model.
  • Dates: pre-orders from 28 October in 35 countries and regions, availability from 6 November.

Where it fits in Apple’s launch week

The iMac is one part of a staggered run of announcements. Apple revealed an OLED iPad mini on 26 October and a 14-inch MacBook Pro with M6 on 27 October. 9to5Mac reports that iMac pre-orders open at 10:10 a.m. Pacific on 28 October.

The M6 chip itself is not new. Apple introduced it on 25 August in the Mac mini, alongside the M5 Ultra in the Mac Studio, which we covered in our M5 Ultra launch report. The iMac with M6 brings that chip to Apple’s family desktop two months later.

What did not change

The 24-inch 4.5K Retina display is the same panel, with 500 nits of brightness and P3 wide colour. The design, the 12MP Center Stage camera, the six-speaker system and the colour-matched accessories also carry over. Engadget points out that Apple still has not delivered a larger 27-inch iMac.

The connectors stay on Thunderbolt 4. The Mac Studio has moved to Thunderbolt 5, which is twice as fast, but the iMac has not.

The M6 Chip Inside the iMac with M6

Upright bicycle floor pump with a pressure gauge, standing for the GPU Neural Accelerators in the iMac with M6

M6 is built on a 2-nanometre manufacturing process, which Apple describes as its first. Smaller transistors pack more logic into the same space and use less power, which is how Apple adds cores without a hotter or thicker machine.

CPU: two more cores and “super cores”

The 12-core version of M6 has 2 super cores, 4 performance cores and 6 efficiency cores. The 10-core version in the cheaper iMac drops one performance core and one efficiency core. Apple says the super cores give M6 the world’s fastest single-threaded CPU performance.

For everyday work Apple claims the iMac with M6 is up to 1.5x faster than the M4 model in apps like Microsoft Excel, and up to 2.5x faster than the iMac with M1. Those are solid gains, but they are a long way short of the 5x AI figure.

GPU with Neural Accelerators

The GPU also gains two cores, to a maximum of 12, and each core now contains a Neural Accelerator. These are small blocks of matrix hardware designed for the maths inside AI models. The M4 GPU in the old iMac had none.

Apple claims up to 2x faster graphics than the M4 iMac for work such as photo editing in Affinity, and up to 2x higher gaming performance with ray tracing. When it launched M6 in August, Apple said the new GPU delivers nearly 30% more peak GPU compute for AI than M5.

The Dual 16-core Neural Engine

The Neural Engine is Apple’s dedicated AI processor. M6 has two of them, each with 16 cores, and Apple says system frameworks can use both at once. The iMac release says this gives “up to 2x faster AI compute than the previous generation”.

That 2x figure is important. It is the Neural Engine on its own, and it is far smaller than the 5x headline. The difference comes from the GPU, as the next section explains.

Memory: up to 170GB/s, with a caveat

Memory bandwidth is how fast the chip can read data from its unified memory. Apple quotes “up to 170GB/s” for the iMac with M6, against 120GB/s for M4. Apple’s August release puts M6 at 10% more than M5.

The words “up to” matter. Coverage of the M6 MacBook Pro by The Gadgeteer reports that its 16GB configuration runs at 153GB/s and only the 24GB and 32GB models reach 170GB/s. Apple’s iMac specification page also says “up to”, so assume the same split until Apple says otherwise.

SpecificationiMac with M4 (October 2024)iMac with M6 (October 2026)
Process3 nm2 nm
CPU8-core or 10-core10-core or 12-core, with 2 super cores
GPU8-core or 10-core10-core or 12-core, Neural Accelerator in each core
Neural Engine16-coreDual 16-core
Memory bandwidth120GB/sUp to 170GB/s
Memory16GB, up to 32GB16GB, up to 32GB (24GB on the 10-core model)
WirelessWi-Fi 6E, Bluetooth 5.3Wi-Fi 7, Bluetooth 6 (Apple N1 chip)
Ethernet optionGigabit2.5Gb
Starting price (US)$1,299$1,499

Unpacking the "5x Faster" AI Claim

Tape measure with its blade pulled out, standing for how Apple measured the 5x AI claim for the iMac with M6

The headline number deserves a close reading, because it describes one kind of AI work. Apple’s wording is precise: “Up to 12x faster on-device AI performance for prompt processing in LM Studio, compared to iMac with M1, and up to 5x faster performance than M4.”

What Apple actually measured

LM Studio is a popular free app for running open-weight language models on a Mac or PC. “Prompt processing” is the first stage of every request. The model reads everything you send it, such as a question, a pasted document or a long chat history, before it writes a single word of its answer.

The second stage is token generation, where the model writes its reply one word fragment at a time. Apple’s 5x claim covers the first stage only. It says nothing about how fast the iMac with M6 writes answers.

Prompt processing is compute-bound

Reading a prompt means running huge batches of matrix maths in parallel. That is exactly the work the new GPU Neural Accelerators were built for, and the M4 GPU lacked them. A 5x gain at this stage is plausible, and it shows up as a shorter pause before the reply starts, especially with long documents.

This matters for real tasks. Summarising a 40-page contract, asking questions of a large spreadsheet export or giving a coding assistant a big file all put heavy load on prompt processing.

Apple’s claimed speed-ups for the iMac with M6 versus the M4 iMac
On-device AI, prompt processing in LM Studio 5x
Graphics, photo editing in Affinity 2x
Gaming with ray tracing 2x
Neural Engine AI compute (versus previous generation) 2x
Productivity, Microsoft Excel 1.5x
Source: Apple press release, 27 October 2026. Bar widths are each multiplier divided by 5. All figures are Apple’s own “up to” claims from its testing.

Token generation is bandwidth-bound

Writing the answer works differently. For each new token, the chip has to read the model’s weights from memory. On a single-user desktop, memory bandwidth sets the ceiling, not compute.

A simple rule of thumb gives that ceiling: bandwidth divided by the size of the model in memory. An 8-billion-parameter model squeezed to 4-bit precision takes roughly 4.4GB, using 0.55GB per billion parameters to allow for overhead. On that arithmetic the iMac with M6 gets about 1.4x more generation headroom than the M4, not 5x.

Theoretical ceiling for writing answers: 8B model at 4-bit (about 4.4GB), tokens per second
iMac with M6, 24GB or 32GB (170GB/s ÷ 4.4GB) about 39
16GB configuration if it runs at 153GB/s (153 ÷ 4.4) about 35
iMac with M4 (120GB/s ÷ 4.4GB) about 27
Our arithmetic from Apple’s bandwidth figures. These are upper bounds; real apps land below them. 170 ÷ 120 = 1.42, so the generation ceiling rises about 42%.

Which configuration Apple compared

Apple’s footnote matters too. The M4 comparison system was the top M4 iMac: 10-core CPU, 10-core GPU, 32GB of memory and a 2TB SSD. That is a fair baseline, since it is the fastest M4 model. But Apple’s “up to” figures normally come from the best M6 configuration, so the $1,499 model with fewer GPU cores will see smaller gains.

None of this means the claim is false. Faster prompt processing is real and useful. It just means the iMac with M6 will feel much faster before an answer starts, and only moderately faster while the answer is being written.

What Local AI Can the iMac with M6 Run?

Garden hose reel with hose coiled on the drum, standing for memory bandwidth when a local AI model writes answers

Apple says the iMac is “a powerful platform for AI while preserving user privacy”. Running a model on the machine keeps the data on the machine, which is the strongest argument for local AI. The practical limit is memory, not speed.

Memory decides model size

A model has to fit in unified memory alongside macOS and your open apps. The chart below uses the same 0.55GB per billion parameters rule for 4-bit models. Treat it as a guide, since real file sizes vary by format and context length.

Approximate memory needed for 4-bit models (billions of parameters × 0.55GB)
70B model: does not fit, even at 32GB about 38.5GB
32B model: needs the 32GB iMac about 17.6GB
14B model: comfortable on 24GB about 7.7GB
8B model: fine on the 16GB base model about 4.4GB
Our arithmetic. Leave roughly 6-8GB for macOS and apps. Bar widths are each size divided by 38.5GB.

The practical takeaway: the 16GB base model suits small assistants and coding helpers. Anyone serious about local models should buy the 12-core model with 32GB, the only iMac with M6 that can hold a 30-billion-parameter class model. For larger AI models, Apple’s answer is the Mac Studio, which our Mac Studio M5 Ultra review covers.

Our earlier piece on 2-bit quantization on a MacBook Pro shows how far aggressive compression can stretch a fixed memory budget, and what it costs in quality.

Apps Apple named

Apple’s release mentions LM Studio Bionic and Topaz Studio as AI apps optimised for Apple silicon, along with Pixelmator Pro and Anthropic’s Claude app. It also mentions “vibe coding”, meaning building software by describing it to an AI model, as a use case.

Several of these run partly or fully in the cloud. Claude’s desktop app, for example, sends requests to Anthropic’s servers, so a faster Neural Engine helps the interface rather than the model. Hybrid tools are a growing category, as our report on Perplexity’s hybrid mode for the Mac explains.

Apple Intelligence and Siri AI on the iMac with M6

The iMac with M6 ships with macOS 27 and the next generation of Apple Intelligence, including Siri AI, which Apple calls “profoundly more capable, conversational, and personal”. Siri AI can search across messages, emails and photos, act across apps and answer questions inside Spotlight. Visual Intelligence gets a keyboard shortcut for asking about anything on screen.

The small print is worth reading. Apple’s footnotes say Siri AI is in beta in macOS 27.2, in English, French, Japanese, Korean, Portuguese and Spanish, and is not available to users under 13. Server-side features carry daily usage limits, and Apple says “expanded access to such features will be available for a fee in the future”. Our guide to the iOS 27.2 Siri AI tools covers the rollout.

Engadget says Apple Intelligence is not yet available in Europe. Apple’s own footnote is narrower: Siri AI is “currently unavailable in the EU in iOS, iPadOS, and watchOS”. It does not list macOS, so check the position in your country before buying for it.

Many Siri AI requests also run on Apple’s servers rather than on the iMac. So the Neural Engine upgrade speeds up on-device features such as writing tools, photo editing and local models, not every Siri answer. Apple’s newest apps lean on this too, as our report on the Creator Studio AI assistant shows.

Price: The iMac with M6 Costs More to Get In

Apple’s release does not mention it, but the entry price has risen. The iMac with M4 started at $1,299 in the US for an 8-core model. That tier is gone. The iMac with M6 starts at $1,499 and £1,499.

ModelUS priceEducation (US)Ports and limits
M4, 8-core CPU and GPU (2024)$1,299$1,249Two Thunderbolt/USB 4, up to 24GB
M4, 10-core CPU and GPU (2024)$1,499$1,399Four Thunderbolt 4, up to 32GB
M6, 10-core CPU and GPU$1,499$1,449Two Thunderbolt 4, up to 24GB and 1TB, one external display
M6, 12-core CPU and GPU$1,699$1,599Four Thunderbolt 4, up to 32GB and 2TB, two external 6K displays

The $1,499 model gave things up

Compare the two $1,499 machines. In 2024, $1,499 bought the 10-core M4 iMac with four Thunderbolt ports and room for 32GB of memory. In 2026, $1,499 buys the 10-core M6 iMac with two Thunderbolt ports, a 24GB ceiling and support for one external display.

To get four ports, 32GB and two external displays, you now pay $1,699. The new chip is faster, but on features the price ladder has moved up by $200 at both ends. The same pattern appeared with the Mac mini with M6, which TechRadar reports starts at $899, $100 more than the M4 model.

Lease and education pricing

Apple Upgrade, a US leasing plan, offers the iMac with M6 from $28.99 a month for 36 months, excluding tax and trade-in credit. That totals $1,043.64 over the lease, and you hand the machine back or upgrade at the end. Education pricing starts at $1,449 and $1,599 in the US and £1,449 and £1,599 in the UK.

Everything Else That Is New

Paint roller resting across a paint tray, standing for faster graphics and photo editing on the new iMac

Beyond the chip, the upgrades are modest but useful, especially for offices and small businesses.

N1 wireless, 2.5Gb Ethernet and faster storage

Apple’s N1 wireless chip brings Wi-Fi 7 and Bluetooth 6, up from Wi-Fi 6E and Bluetooth 5.3. The optional Ethernet port moves to 2.5Gb, which Apple says gives up to 2.5x faster wired connections. Apple also claims up to 2x faster SSD reads and writes than the M4 model.

For a business, the Ethernet change is the most practical. Moving large files to a network drive or a NAS (network-attached storage) box is limited by the wire, and 2.5Gb is now common on office switches.

Display, camera and ports

The 4.5K display supports a nano-texture glass option on the 12-core model. The 12-core iMac with M6 can drive two external 6K displays, which Apple says gives more than 50 million pixels in total. The 12MP Center Stage camera with Desk View and the three-microphone array are unchanged.

Environment

Apple says the new iMac uses 100% recycled aluminium in the stand, 100% recycled rare earth elements in all magnets, and is made with 40% renewable electricity across the supply chain. The packaging is 100% fibre-based.

iMac with M6 vs the Alternatives

The iMac is not the only way to get M6. The table below compares Apple’s M6 desktops and laptop on price and the trade-offs that matter for AI work.

MachineStarting price (US)Best forMain trade-off
Mac mini with M6$899 (16GB, 256GB)Cheapest M6, bring your own screenDisplay, keyboard and mouse cost extra
iMac with M6, 10-core$1,499 (16GB, 256GB)Home and front-desk useTwo ports, 24GB maximum
iMac with M6, 12-core$1,699 (16GB, 256GB)All-in-one with room for local AI32GB maximum, Thunderbolt 4 only
14-inch MacBook Pro with M6$1,999 (16GB, 1TB)Portable work and travelHigher price, smaller screen

Against a Windows AI PC

Microsoft and its hardware partners are pushing Copilot+ PCs with their own neural processors, as we covered in Microsoft’s AI PC push against Apple. Windows all-in-ones are often cheaper, but few match the iMac’s display or its unified memory, which lets the GPU use most of the system RAM for models.

Apple’s own comparison claims the iMac display has 5x more resolution and nearly 70% more brightness than the bestselling 24-inch all-in-one PC. That is a display claim, not an AI one, and Apple does not name the PC.

Who Should Upgrade to the iMac with M6

Hand trolley carrying stacked moving boxes, standing for upgrading from an Intel or M1 iMac

The answer depends on what you own now. Apple’s comparisons give a useful guide, as long as you remember they are “up to” figures from Apple’s own testing in September and October 2026.

Intel iMac owners

This is the strongest case. Apple claims up to 18x faster performance than the most popular Intel-based iMac. When it launched the M4 iMac, Apple put the same comparison at 6x, so the 18x figure most likely reflects an AI-heavy test rather than everyday work. Either way, Intel Macs cannot run Apple Intelligence at all, and macOS support for them is ending.

M1 iMac owners

The iMac with M1 is now five years old. Apple claims up to 2.5x faster productivity, 3.3x faster graphics and 12x faster prompt processing. If you want local AI or Siri AI’s on-device features, the upgrade is worth it. If you mainly browse and write, the M1 still copes.

M4 iMac owners

Skip it unless local AI is central to your work. The everyday gain is about 1.5x, and the token generation ceiling rises by about 40%. The M4 iMac already runs Siri AI and Apple Intelligence.

Apple’s claimed AI prompt-processing speed-up for the iMac with M6, by the machine you own now
Versus iMac with M1 12x
Versus iMac with M4 5x
Source: Apple press release, 27 October 2026, LM Studio prompt processing. Bar widths are each multiplier divided by 12. 12 ÷ 5 = 2.4, which implies the M4 iMac itself ran this test about 2.4x faster than the M1.

What It Means for Businesses Running Local AI

For small offices, the iMac with M6 makes a strong front-desk or design machine. But the AI story needs care before a fleet purchase.

First, decide which AI work truly has to stay on the device. Client records, contracts and health data are good candidates for local models. General research and drafting can run in cloud services with business contracts and data controls.

Second, buy the memory you need at the start. Unified memory cannot be upgraded later, so a 16GB iMac bought for local AI in 2026 will stay a 16GB iMac. For teams running models, the 32GB 12-core model is the sensible floor.

Third, plan how you will manage the machines. Apple Intelligence settings, Siri AI data access and third-party AI apps should be controlled centrally through a mobile device management tool, not left to each user. Our device management service covers Mac fleets, and our AI strategy work helps decide which tasks belong on local models.

Finally, match the AI to the job. Most staff will get more from well-chosen cloud AI tools than from running natural language processing models on their own desks. Local AI earns its place where privacy, offline use or cost per request matter most.

Our Verdict

The iMac with M6 is a real upgrade, and the “huge leap” is not empty marketing. Neural Accelerators in the GPU, a Dual 16-core Neural Engine and a 2-nanometre chip make long prompts and on-device AI features much faster, and Wi-Fi 7 and 2.5Gb Ethernet are welcome.

But the 5x figure describes prompt processing, the part of local AI that benefits most from new compute. Writing answers is limited by memory bandwidth, which rises by about 40%. And the price of entry has gone up by $200, with the $1,499 model losing ports and memory headroom compared with its predecessor.

For Intel and M1 owners, the iMac with M6 is an easy recommendation. For local AI, buy the 12-core model with 32GB. For M4 owners, it is a nice upgrade, not a necessary one.

iMac with M6 FAQs

When can I buy the iMac with M6?

Pre-orders open on 28 October 2026 on apple.com and in the Apple Store app in 35 countries and regions, including the US and UK. Deliveries and in-store availability begin on Friday 6 November.

How much does the iMac with M6 cost?

It starts at $1,499 or £1,499 for the 10-core CPU and GPU model, and $1,699 or £1,699 for the 12-core model. Both start with 16GB of memory and a 256GB SSD. Education pricing starts at $1,449 and £1,449.

Is the iMac with M6 really 5x faster at AI?

Apple claims up to 5x faster on-device AI than the M4 iMac for prompt processing in LM Studio, which is the stage where a model reads your input. Generating the answer is limited by memory bandwidth, which rises from 120GB/s to up to 170GB/s.

How much memory does the iMac with M6 support?

The 10-core model can be configured with 16GB or 24GB. The 12-core model offers 16GB, 24GB or 32GB. Memory cannot be upgraded after purchase.

Does the iMac with M6 have a bigger screen?

No. It keeps the 24-inch 4.5K Retina display with 500 nits of brightness. A nano-texture glass option is available on the 12-core model. Apple has not announced a larger iMac.

Does it run Siri AI?

Yes. The iMac with M6 ships with macOS 27 and the next generation of Apple Intelligence, including Siri AI in beta in macOS 27.2. Some features have daily usage limits and language and regional restrictions.

References