Owners of M4 Macs may feel the temptation to upgrade to the M6 Mac mini, MacBook Pro, or iMac. Here's what to expect with the jump to Apple's newest silicon.
In August 2026, Apple introduced the M6 as its new entry-level chip, as well as the M6 generation itself. Brought out with the Mac mini, the M6 was framed by Apple as a way to benefit more from AI, with changes that made it an ideal chip for machine learning workloads.
The introduction took place just over two years after Apple launched its first M4 chip in the iPad Pro. The M6 is also a two-generation jump for some models, such as the Mac mini, which skipped the base M5 chip completely.
With Apple bringing out more hardware using M6, more consumers are going to look at the new hardware and seriously consider upgrading from their M4 equivalents.
This is the breakdown of what Apple's M6 chip provides, and how it is different from the M4.
M6 vs M4: Specifications
| Specifications | M6 | M4 |
|---|---|---|
| Process | 2 nanometer | 3 nanometer |
| Cores | 2 Super, 4 Performance, 6 Efficiency | 4 Performance, 6 Efficiency |
| Neural Engine | Dual 16-core | 16-core |
| GPU | 12-core, Hardware-accelerated ray-tracing, Neural Accelerators | 8, 9, 10-core, Hardware-accelerated ray-tracing |
| Media Engine | Hardware-accelerated H.264, HEVC, ProRes, and ProRes RAW Video decode engine, Video encode engine, ProRes encode and decode engine, AV1 decode | Hardware-accelerated H.264, HEVC, ProRes, and ProRes RAW Video decode engine, Video encode engine, ProRes encode and decode engine, AV1 decode |
| Memory Support | 16GB, 24GB 32GB | 8GB, 12GB, 16GB, 24GB, 32GB |
| Memory Bandwidth | 170GB/s | 120GB/s |
| Transistors | 46 billion | 28 billion |
M6 vs M4: Processes and transistors
The M6 and M4 are made by Apple's usual chip foundry partner, TSMC. However, they are made using two different processes.
The M4 was made using a 3-nanometer process, while the M6 uses a 2-nanometer version.
There are usually quite a few benefits for a so-called die shrink, with the first being density. You can fit more transistors in a smaller process level than a larger one, simply because everything is much smaller.
When it comes to transistors, TSMC was able to squeeze 46 billion onto the M6 die. The M4, meanwhile, has just over half that figure at 28 billion transistors.
M6 vs M4: Die shrinks let you fit transistors more densely.
The more transistors on a chip, the more potential calculations and processes it can perform.
The smaller dies can also get advantages in other areas, such as requiring less power for processing. Heat dissipation is also improved on smaller processes.
The die shrink means that the M6 can run cooler and with less energy than the M4 on comparable workloads.
M6 vs M4: Cores and clock speeds
Modern processors don't just have multiple cores, but multiple core types, too. Cores at different clock speeds can help a system better manage power in relation to performance needs.
A high-performance core may not be used while a system is idling, leaving background work for lower-powered efficiency cores to do. When it comes to completing a task, those performance cores come into play.
The M4 has ten cores in Macs, consisting of four performance cores, and six efficiency cores. A nine-core CPU was also available in the iPad Pro, with three performance cores.
The performance cores are clocked at 4.4GHz, while the efficiency cores run at 2.85GHz.
When it comes to the M6, Apple decided to add a new core type to the equation. Alongside the four performance and six efficiency cores, there were a pair of "super" cores that perform at an even higher level.
In the first M6 release, this meant the super cores ran at 4.79GHz, the performance cores at 4.5GHz, and the efficiency cores at 2.94GHz.
Not only does the M6 run at faster clock rates per-core overall, but there's also the benefit of having two extra cores available.
M6 vs M4: Memory
One of the few advantages the M4 has over the M6 is memory quantity support, at least in theory.
The M4 has been used in devices with configurations between 8GB and 32GB of memory, while the M6 goes from 16GB to 32GB allocations.
The reason for this is that there were iPad Pro models with M4 that had just 8GB of memory. There hasn't been an M6 device with less than 16GB.
M6 vs M4: M6 arrived with the Mac mini. There's not been an M6 iPad Pro yet, at the time of reporting.
That said, the M6 still wins out when it comes to bandwidth.
The M4 has a memory bandwidth maximum of 120GB/s. That's dwarfed by the M6, which goes to 170GB/s.
Memory bandwidth is essential, since Apple relies on using Unified Memory. The same memory is used by the CPU, GPU, and other components to minimize data duplication, which is common when separate pools of memory are being used.
M6 vs M4: Graphics and the Neural Engine
While in previous years, we would separate the GPU and the Neural Engine, changes in Apple's hardware landscape means they should be discussed together.
First, the Neural Engine in the M4 has 16 cores, as is tradition, with the capability of performing up to 38 trillion operations per second.
By comparison, the Neural Engine in the M6 is a "Dual 16-core" version. This means it has 32 cores to play with, double that of the M4.
On the GPU side, the M4 has a few configurations depending on the device, at eight, nine, or ten cores. The GPU also has hardware-accelerated ray-tracing, which helps with rendering graphics.
The M6, meanwhile, has a 12-core GPU in its base configuration. The newer GPUs are always declared as being faster than the previous generations, but the M6 has one more advantage over the M4.
The M6 GPU cores have Neural Accelerators in each core. Introduced with the M5, they handle matrix-multiplication operations, which are a bedrock of processing AI and large language models (LLMs).
While the M6 benefits from double the Neural Engine cores for a start, the GPU's Neural Accelerators gives it a massive advantage when it comes to AI processing in general.
That makes the M6 a quite desirable chip for those working in AI.
M6 vs M4: Benchmarks
A good way to demonstrate the performance difference between the M4 and M6 chips is by benchmarking them on the same sort of hardware. Since the Mac mini exists in both M4 and M6 variants, it's an ideal vehicle for this.
M6 vs M4: Geekbench single-core
Under Geekbench 7's listings, the M4 Mac mini manages a single-core score of 3,277. The super core of the M6 Mac mini helps push it further, right up to 4,002, a 22% improvement.
M6 vs M4: Geekbench multi-core
The multi-core score of the M4 Mac mini is decent, with the ten cores managing 15,304 when working together. However, the faster cores and the increased count in the M6 Mac mini reach 21,881, over 42% better.
M6 vs M4: Geekbench Metal
Under Metal, there's a similar gulf of performance. The M4's 54,227 is heavily beaten by the 89,375 of the M6, a 64% improvement.
M6 vs M4: Two generations of improvement
Looking at the M6 against the M4 was always going to be a performance bloodbath. There is a lot of change that happens between two chip generations, with every small alteration helping to grow the gap.
The M6 certainly is an AI-forward chip, with masses of graphical performance aiding the double-sized Neural Engine. And then there's the CPU core improvements and the additional two Super cores to account for, too.
It's easy to see how much of an improvement the M6 has, which directly translates into everyday tasks. You can certainly see where Apple is going with its future Apple Silicon efforts.
That all said, the M4 is still very capable. It's not what you would consider slow, but the M6 is helping show its age a bit.
