Apple’s new M6 chip has been shipping inside the redesigned Mac mini since September 22, and the first head-to-head numbers against the year-old M5 are now public. Between Apple’s own performance claims, a leaked Geekbench 7 result first reported by MacRumors, and a hands-on comparison from Engadget, a clearer picture of the M6-vs-M5 gap has emerged, and it is smaller than the marketing slides suggest.
Apple announced the M6 on August 25, 2026, debuting the chip in a refreshed Mac mini that starts at $899 ($799 for education buyers). That is the same launch window Apple used for the M5 a year earlier, when the 14-inch MacBook Pro introduced that chip on October 15, 2025. The two launches bookend a year in which Apple also shipped M5 Pro, M5 Max, and M5 Ultra variants, and reportedly set up the entire Pro-tier lineup to skip the M6 generation altogether, as detailed in our earlier report on Apple reportedly skipping M6 Pro, Max, and Ultra chips for M7 in 2027.
For anyone deciding between a new M6 Mac mini and an M5 MacBook Pro still on store shelves, the practical question is simple: how much performance are you actually gaining, and is it worth the price difference? Here is what the numbers, the leaks, and Apple’s own claims say.
What Apple Actually Announced on August 25
Apple’s M6 debut followed a pattern the company has used for every silicon generation since the M1: a base chip ships first in the cheapest machine in the lineup, with Pro, Max, and Ultra variants expected to follow. This time the base chip landed in the Mac mini, not a MacBook, and it arrived alongside refreshed M5 Pro Mac mini configurations and new M5 Max and M5 Ultra Mac Studio models, all launching on the same day, September 22, 2026, according to MacRumors’ launch coverage.
That lineup split is the first sign that Apple M6 vs M5 is not a clean generational swap. The M6 only replaces the base M5 tier for now. Everything above it, Pro, Max, and Ultra, is still running M5-generation silicon, which is why shoppers comparing a baseline Mac mini to a higher-end MacBook Pro are effectively comparing two different chip families rather than two steps on the same ladder.
Inside the Apple M6: A 12-Core Redesign
The headline change in the M6 is the core count. Apple’s CPU now carries 12 cores split across two super cores, four performance cores, and six efficiency cores, two more total cores than the M5’s 10-core layout. The GPU side also grew to 12 cores, and Apple says every GPU core now includes a dedicated Neural Accelerator, the same architecture it introduced with the M5 generation but scaled up.
Apple’s official line is that the M6 delivers the world’s fastest single-threaded performance of any chip it has shipped, and the company backs that with a specific multithreaded comparison: up to 1.2x faster multithreaded performance than the M5, and up to 2.4x faster than the original M1. Those are Apple’s own figures from its August 25 announcement materials, not third-party benchmarks, and the gap between “1.2x faster” and the double-digit AI claims circulating elsewhere is worth sitting with for a moment, because it is the most honest number Apple has published about raw CPU throughput in this generation.
Engadget’s hands-on comparison, published the same day as Apple’s announcement, adds a detail Apple’s own materials gloss over: the M6 also claims a 30 percent AI performance boost over the M5, driven by a new Dual 16-core Neural Engine working alongside those per-core GPU Neural Accelerators. That AI-specific number is considerably larger than the CPU multithreading gain, which tells you where Apple actually spent its transistor budget this generation.
Apple M5 Recap: The Chip M6 Has to Beat
The M5 debuted on October 15, 2025, inside the 14-inch MacBook Pro, and it is still the chip most new Mac buyers will encounter if they are not specifically shopping for a Mac mini. It carries a 10-core CPU (four super cores and six efficiency cores), a 10-core GPU with Neural Accelerators and hardware-accelerated ray tracing, and a 16-core Neural Engine. In its cited MacBook Pro configuration, it supports 153GB/s of unified memory bandwidth.
Apple’s claims for the M5 at launch were aggressive on the AI side specifically: up to 3.5x faster AI performance than the M4, and up to 6x faster than the M1. Those figures are not directly comparable to the M6’s CPU-multithreading claim, since Apple measured different workloads for each chip’s launch, but they explain why the M5 still holds up reasonably well against a newer chip that mostly added CPU and GPU cores rather than reworking the Neural Engine pipeline from scratch.
Apple M6 vs M5: Full Spec Comparison
Here is how the two chips line up on paper, using Apple’s own published specifications plus the process-node and memory-bandwidth detail Engadget confirmed in its launch-day report.
| Spec | Apple M6 | Apple M5 |
|---|---|---|
| Announced | August 25, 2026 | October 15, 2025 |
| Debut product | Mac mini | 14-inch MacBook Pro |
| CPU cores | 12 (2 super + 4 performance + 6 efficiency) | 10 (4 super + 6 efficiency) |
| GPU cores | 12-core, Neural Accelerator per core | 10-core, Neural Accelerator per core, ray tracing |
| Neural Engine | Dual 16-core (per Engadget) | 16-core |
| Manufacturing process | TSMC 2nm (per Engadget, MacRumors) | TSMC 3nm (per Engadget) |
| Unified memory bandwidth | Up to 170GB/s (per Engadget); 153GB/s on 16GB config | 153GB/s (14-inch MacBook Pro config) |
| Apple’s CPU claim | Up to 1.2x faster multithreaded vs. M5 | Baseline for that comparison |
| Apple’s AI claim | Up to 30% faster AI vs. M5 (per Engadget) | Up to 3.5x faster AI vs. M4 |
| Starting price | $899 Mac mini ($799 education) | $1,999 as listed in CNET’s October 2026 buying guide |
Single-Core vs Multithreaded: What “20% Faster” Actually Means
Apple’s 1.2x multithreaded claim is a company-reported figure, and it typically holds up reasonably well in third-party testing, but it is not the only number circulating. MacRumors reporter Joe Rossignol flagged the first unverified Geekbench 7 result for the M6 shortly after launch, and the gap it shows is noticeably wider than Apple’s own framing.
According to that Geekbench 7 listing, the M6 scored 4,071 on single-core and 22,783 on multi-core. MacRumors compared that against the M5’s average database scores of 3,641 single-core and 17,948 multi-core, calculating roughly a 12 percent single-core gain and roughly a 27 percent multi-core gain for the M6, both well above the raw percentage Apple’s 1.2x figure implies on paper. MacRumors also noted that the M6’s 4,071 single-core score tops every other Mac and PC currently in the Geekbench 7 database, including the M5 Max (3,731) and Qualcomm’s Snapdragon X2 Elite Extreme (3,396).
It is worth flagging, as MacRumors itself did, that this is a single unverified benchmark run, not a broad sample. Early Geekbench listings for new Apple silicon have occasionally run hot or cold relative to shipping-unit averages, so treat the 27 percent multi-core gain as directionally interesting rather than a locked-in figure. What it does confirm is that the real-world gap between M6 and M5 is likely closer to “noticeably faster” than the conservative 1.2x Apple chose to put on a slide.
The Process Node Question: 2nm Changes the Math
The most consequential change in the M6 is not the core count, it is the manufacturing process underneath it. Engadget’s launch-day report states plainly that the M6 is Apple’s first chip built on TSMC’s 2nm process, a jump from the 3nm node that defined the M3, M4, and M5 generations. That node shift is also why Apple can claim smaller transistors are driving a meaningful chunk of this generation’s efficiency and performance gains, independent of the extra cores.
That earlier node transition is also showing up in the supply chain. As we reported separately, DRAM now costs more per chip than TSMC’s 2nm silicon itself, a dynamic that squeezes margins on exactly the kind of memory-bandwidth upgrades Apple is advertising for the M6. Apple has not disclosed M6 wafer pricing, and nothing in the company’s public materials confirms a specific cost premium for the new node, so treat any claims about M6’s production cost as speculation until Apple or TSMC says otherwise.
Memory Bandwidth: Bigger Number, Smaller Catch
Apple advertises up to 170GB/s of unified memory bandwidth for the M6 in its higher-memory configurations, a roughly 10 percent increase over the M5’s 153GB/s, according to Engadget’s breakdown of the spec sheet. The catch is in the word “up to.” The base 16GB M6 Mac mini configuration is listed at 153GB/s, identical to the M5’s figure, meaning the bandwidth bump only shows up once you pay for more memory.
Engadget’s review also notes that the base RAM configuration is configurable up to 32GB, which is a meaningful ceiling for a chip aimed at “everyday” users rather than professionals running Final Cut Pro renders or large local AI models. That framing matters for anyone comparing the M6 Mac mini against Apple’s own higher-end silicon, including the 36-core M5 Ultra, which supports 1.2TB/s of memory bandwidth, more than seven times what the base M6 offers even at its best configuration.
Pricing Breakdown: Mac Mini M6 vs MacBook Pro M5
The M6 Mac mini starts at $899 in the United States, or $799 for verified education buyers, making it Apple’s cheapest entry point into current-generation silicon. The M5, by contrast, only ships in the MacBook Pro line, which carries a far higher price floor. CNET’s October 2026 buying guide lists the 14-inch M5 MacBook Pro at $1,999, more than double the M6 Mac mini’s starting price, though that figure reflects a specific configuration rather than Apple’s original base launch price.
| Machine | Chip | Starting price | Launch date |
|---|---|---|---|
| Mac mini (M6) | Apple M6 | $899 ($799 education) | Sept. 22, 2026 |
| Mac mini (M5 Pro) | Apple M5 Pro | Not confirmed in Apple’s public pricing pages | Sept. 22, 2026 |
| Mac Studio (M5 Max / M5 Ultra) | Apple M5 Max / M5 Ultra | Not confirmed in Apple’s public pricing pages | Sept. 22, 2026 |
| 14-inch MacBook Pro (M5) | Apple M5 | $1,999 (per CNET, Oct. 2026) | Oct. 15, 2025 |
The practical takeaway is that M6 and M5 are not really competing for the same buyer. Someone shopping at $899 is choosing a desktop machine over a laptop, and someone paying close to $2,000 for a MacBook Pro is paying for portability, a display, and a battery, not just a faster chip. Treating this as a pure chip-for-chip value comparison undersells how different the two machines actually are.
What’s Confirmed vs What’s Still Rumor
Several numbers attached to the M6 online are more solid than others. Apple’s own claims, the 12-core CPU and GPU counts, the $899 starting price, and the “up to 1.2x” multithreaded figure, come directly from the company’s August 25 announcement and are not in dispute. The 2nm process, the 170GB/s bandwidth ceiling, the Dual 16-core Neural Engine, and the 30 percent AI performance claim all trace back to Engadget’s launch-day reporting rather than Apple’s own published spec sheet, which is a meaningfully different kind of source but still a named, on-the-record one.
The Geekbench numbers sit in a third category entirely: a single leaked, unverified result that MacRumors chose to report because it was the first data point available, not because it has been confirmed across multiple test units. Readers should weigh that 27 percent multi-core gain accordingly, as an early signal rather than a settled benchmark average.
Apple’s Silicon Roadmap: Why Pro, Max, and Ultra Skipped M6
The most structurally interesting part of this launch is what did not happen. Apple did not refresh its Pro, Max, or Ultra tiers with M6 silicon. Instead, the September 22 Mac Studio refresh uses M5 Max and M5 Ultra chips, and the higher-end Mac mini uses an M5 Pro chip, both still built on the 3nm process that defined the entire M5 family. Our earlier reporting detailed how Apple reportedly plans to skip M6 Pro, Max, and Ultra variants entirely, pushing that tier straight to M7 in 2027.
That staggered rollout means the M6 vs M5 comparison most people are searching for is really a comparison between Apple’s newest low-end chip and its one-year-old high-end chips, which is backwards from how most buyers assume Apple’s naming works. Someone who wants the fastest Apple silicon available in October 2026 is not buying an M6 at all, they are buying the M5 Ultra inside the Mac Studio, a chip Apple is still selling as its current flagship more than a year after its underlying architecture debuted.
How M6 Stacks Up Against Windows-on-ARM and x86 Rivals
Apple is not the only company making performance-per-watt claims this year. The M6’s reported 4,071 Geekbench single-core score, per MacRumors’ tracking, already tops Qualcomm’s Snapdragon X2 Elite Extreme at 3,396, a chip we covered in detail when it went up against Intel’s Core Ultra lineup earlier this year. That gap gives Apple a real talking point against Windows-on-ARM machines, even though the M6 currently only ships in a desktop, not a laptop that would compete directly with Snapdragon-powered ultrabooks.
On the Windows side, machines built around last-generation x86 and ARM silicon are increasingly being marketed on price rather than raw performance. The Machenike 14 Ultra, a $1,267 Windows laptop positioned as a MacBook Pro rival, is a good example of that pricing pressure, undercutting the $1,999 M5 MacBook Pro by more than $700 while targeting similar creative and productivity workloads. Apple’s silicon advantage is real, but it is no longer the only variable buyers are weighing.
Five Years of Apple Silicon: From M1 to M6
Apple’s self-reported multipliers give a rough sense of how far the line has come since the M1 launched in late 2020. The M5 claimed up to 6x faster AI performance than the M1, and the M6 claims up to 2.4x faster multithreaded CPU performance than that same baseline. Those are two different metrics measuring two different things, AI throughput versus general multithreading, so they should not be read as a single continuous curve, but both numbers point to the same broader story: Apple has leaned harder on AI-specific silicon (Neural Engines, per-core GPU accelerators) in the last two generations than on raw CPU core scaling.
The M6’s jump to a 2nm process is the first manufacturing-node change since the M3 generation moved to 3nm back in 2023, per Engadget’s reporting. That makes M6 the first “new node” Apple chip in roughly three years, which is part of why its real-world gains, even the modest 1.2x multithreaded figure, are being watched closely as an early signal for how much further TSMC’s 2nm process can scale once Apple’s Pro, Max, and Ultra chips eventually move to it.
Market Reaction and Analyst Takeaways
Early coverage has been notably measured compared to past Apple silicon launches. Engadget’s own headline framing, that M5 owners “shouldn’t feel any FOMO,” captures the general tone: this is being read as a solid, expected generational step rather than a must-upgrade moment. That is a shift from the M1 and M4 launches, which were covered as larger architectural leaps.
Part of that muted reaction comes from the lineup split itself. Because the fastest machines Apple sells right now, the M5 Max and M5 Ultra Mac Studios, are not running M6 silicon at all, the chip’s debut reads more like a quiet refresh of Apple’s cheapest desktop than a flagship event. That is a deliberate sequencing choice on Apple’s part, not an accident, and it mirrors how the company staged the M1 generation: entry-level silicon first, Pro and Max variants months later, Ultra last.
What This Means for Mac Buyers Right Now
If you need a Mac today and your budget tops out near $900, the M6 Mac mini is the newest, most efficient chip Apple sells at that price, full stop. If you need a laptop and your budget is closer to $2,000, the M5 MacBook Pro remains the only current option, since M6 has not reached any MacBook yet. And if raw performance is the only thing that matters to you, the M5 Max and M5 Ultra Mac Studio configurations still outrun the base M6 by a wide margin, according to MacRumors’ own multi-core leaderboard, which placed the M5 Ultra at 52,516 and the base M6 at 22,783 in multi-core Geekbench 7 testing.
For current M5 MacBook Pro owners specifically, there is no M6 MacBook to upgrade to yet, and Apple has given no public timeline for one. Waiting makes sense only if you were already planning to wait; nothing about this launch changes the calculus for someone who bought an M5 machine in the past year.
Predictions: Where Apple Silicon Goes Next
- An M6 MacBook Air and M6 MacBook Pro are likely within the next six to twelve months, following Apple’s usual pattern of moving a new base chip from desktop to laptop, though Apple has not confirmed a date.
- M6 Pro, Max, and Ultra variants are unlikely to ship in 2026 at all, based on our earlier reporting pointing to a 2027 arrival tied to the M7 generation.
- Expect more unverified Geekbench leaks like the one MacRumors reported to keep surfacing as M6 reaches more devices, gradually narrowing the gap between Apple’s conservative 1.2x claim and the benchmark community’s real-world averages.
- TSMC’s 2nm process, now proven in a shipping Apple chip, will likely show up in competing silicon from Qualcomm and others within the next one to two product cycles, intensifying the single-core race MacRumors’ leaderboard already shows tightening.
- Pricing pressure from sub-$1,300 Windows machines like the Machenike 14 Ultra will likely push Apple to defend the M5 MacBook Pro’s price point with software and ecosystem arguments rather than chip specs alone.
Frequently Asked Questions
Is the Apple M6 chip out yet?
Yes. Apple announced the M6 on August 25, 2026, and the Mac mini with M6 began shipping on September 22, 2026, starting at $899.
What is the main difference between Apple M6 and Apple M5?
The M6 has a 12-core CPU and 12-core GPU versus the M5’s 10-core CPU and 10-core GPU. Apple also moved the M6 to a TSMC 2nm manufacturing process, up from the 3nm process used in the M5, per Engadget’s reporting, and Apple claims up to 1.2x faster multithreaded CPU performance than the M5.
Does the Apple M6 come in a MacBook Pro?
Not yet. As of this writing, the M6 ships only in the Mac mini. The M5 remains the current chip in the 14-inch MacBook Pro line, alongside M5 Pro and M5 Max configurations.
Is the M6 Mac mini worth upgrading from an M5 machine?
For most M5 owners, no. Apple’s own performance claims put the generational gain at roughly 20 percent in multithreaded CPU work, and Engadget’s coverage framed the M6 as a chip M5 owners should not feel pressured to chase. The upgrade makes more sense for anyone moving up from an M1, M2, or M3-era Mac.
How much faster is the M6 than the M5 in real benchmarks?
An early, unverified Geekbench 7 result reported by MacRumors showed the M6 scoring about 12 percent higher on single-core and about 27 percent higher on multi-core compared to average M5 scores. Those figures come from a single leaked test, not a broad sample, so they should be treated as preliminary.
What manufacturing process does the Apple M6 use?
According to Engadget’s launch-day reporting, the M6 is built on TSMC’s 2nm process, Apple’s first chip on that node. The M3, M4, and M5 were all built on TSMC’s 3nm process.
When will M6 Pro, Max, and Ultra chips launch?
Reports indicate Apple may skip Pro, Max, and Ultra variants of the M6 entirely, moving that tier directly to the M7 generation expected in 2027. The current Mac Studio and higher-end Mac mini configurations instead use M5 Pro, M5 Max, and M5 Ultra chips, launched September 22, 2026.
Should I buy the M5 MacBook Pro now or wait for an M6 MacBook Pro?
If you need a laptop today, the M5 MacBook Pro, listed at $1,999 in CNET’s October 2026 buying guide, is the only current option, since Apple has not announced an M6 MacBook of any kind or given a timeline for one.
Related
- Apple Reportedly Skips M6 Pro/Max/Ultra for M7 in 2027
- Apple’s M5 Ultra Brings 36 Cores, 1.2TB/s Bandwidth
- M5 Ultra Mac Studio Review: 90% Faster, $12,299
- Snapdragon X2 Beats Intel Core Ultra by 31%, Costs $750 Less
- Machenike 14 Ultra: $1,267 Rival to the MacBook Pro
- DRAM Now Costs 54% More Than TSMC 2nm Silicon
