Apple M6 12 Cores
Apple M6 12 Cores: Early Benchmarks and M5 Comparison
Apple M6 12 Cores is Apple’s first 2nm processor, pairing a 12-core CPU with a 12-core GPU. Unified memory bandwidth has increased to 170 GB/s. In early testing, the M6 scores around 4,500 points in Geekbench 6 Single and more than 21,000 in Multi. Retail M6 systems are not yet available, so these results should still be treated as preliminary.
What Changed Compared to M5
The M6 CPU includes two Super Cores, four Performance Cores, and six Efficiency Cores, for a total of 12 cores versus 10 on the M5.
The M6 is manufactured on a 2nm process, compared with 3nm for the M5.
Unified memory bandwidth increased from 153 to 170 GB/s. Maximum memory capacity remains unchanged at 32 GB.
The GPU now has 12 cores, up from 10 on the M5.
Preliminary Benchmark Results
Retail M6 systems are not yet available. The M6 results in the table therefore come from an early non-commercial sample and may differ from scores recorded on production hardware.
| Chip | Geekbench 6 Single | Geekbench 6 Multi | Cinebench 2024 Single | Cinebench 2024 Multi |
|---|---|---|---|---|
| Apple M6 12-Core* | 4501 | 21312 | 220 | 1368 |
| Apple M5 10-Core | 4276 | 17862 | 208 | 1183 |
| Apple M4 10-Core | 3777 | 14818 | 172 | 982 |
| Apple M4 Pro 14-Core | 3899 | 22639 | 172 | 1678 |
* M6 results are from an early non-commercial sample.
In Geekbench 6 Single, the M6 outperforms the M5 by about 5% and the M4 by 19%.
In Geekbench 6 Multi, the M6 is about 19% faster than the M5 and 44% faster than the M4.
In Cinebench 2024 Multi, the M6 leads the M5 by roughly 16% but remains about 18% behind the M4 Pro 14-Core.
The M6 comes close to the M4 Pro in Geekbench 6 Multi, but that does not mean the two chips deliver the same performance under sustained workloads. The M4 Pro has more performance cores, which gives it an advantage in longer multi-threaded tasks.
12-Core GPU and AI
The M6 uses a 12-core GPU with updated shader blocks, Dynamic Caching, and hardware ray tracing.
Apple claims nearly 30% higher peak GPU performance in AI workloads compared with the M5, along with a 50% increase in geometry processing performance.
Each GPU core includes a Neural Accelerator for supported AI operations. The M6 also features a Dual 16-core Neural Engine.
For media workloads, the M6 supports hardware acceleration for H.264, HEVC, ProRes, and ProRes RAW, as well as hardware AV1 decoding.
The First Mac with Apple M6
The first computer based on the M6 is the Mac mini. The base configuration includes a 12-core CPU, a 12-core GPU, 16 GB of unified memory, and a 256 GB SSD.
Sales of the M6 Mac mini begin on September 22, 2026, so there is not yet enough data to evaluate sustained performance on retail systems.
Where M6 Falls Short of Pro Chips
Despite the increase in multi-threaded performance, the M6 remains the base M-series chip.
The 32 GB unified memory limit may become restrictive for large local AI models and workloads that require substantial amounts of memory.
Pro chips also have more performance CPU cores and faster GPUs. As a result, they remain ahead in sustained multi-threaded workloads even when the M6 comes close in individual benchmarks.
Conclusion
The biggest improvement over the M5 is in multi-threaded performance. In early Geekbench 6 testing, the M6 is about 19% faster in Multi, while the Single score improves by around 5%.
In Geekbench 6 Multi, the M6 comes close to the M4 Pro 14-Core, although the Pro chip still has an advantage in more demanding sustained workloads.
For now, the clearest reference point is the roughly 19% lead over the M5 in Geekbench 6 Multi. Tests of retail Mac mini systems after September 22 will show how closely production hardware matches these early results.
CPU Comparison
Compare Apple M6 12 Cores with other CPUs
Basic
CPU Specifications
Memory Specifications
GPU Specifications
AI Specifications
Interfaces and Ports
Benchmarks
Compared to Other CPU
Share in social media
Or Link To Us
<a href="https://cputronic.com/cpu/apple-m6-12-cores" target="_blank">Apple M6 12 Cores</a>