Apple’s new generation M6 processor is attracting attention in the technology world by surpassing its predecessor M4 with revolutionary improvements in artificial intelligence workloads. Current tests reveal that the new Mac mini models with the M6 ​​chip offer much higher efficiency and speed compared to the M4, especially in prompt processing, token generation, and visual rendering processes.

This performance increase achieved in just one generation demonstrates the tangible effects of Apple’s transition to 2nm architecture on artificial intelligence applications. Users can now run much more complex models locally while simultaneously enjoying a cooler working environment with lower power consumption.

The M6 ​​​​Processor Raises the Bar in AI Workloads

Extensive tests conducted by YouTuber Alex Ziskind prove that the M6 ​​​​architecture represents a significant leap forward, especially for users working with local AI models. Comparisons made using the Qwen3.5-9B model show that the M6 ​​chip reaches a speed of 742 kT/s during the prompt process.

Considering that the M4 processor remains at the 210 kT/s limit, this indicates a massive performance difference of 253.3%.

The data highlights the critical role of the new neural accelerators that Apple has integrated into each of its 12 GPU cores in this achievement.

Memory Capacity Allows New Models to Run

The M6 ​​​​Mac pushes the limits of the previous generation with its supported 32GB combined memory capacity. This hardware enhancement means that memory-intensive models such as the Qwen3.8-27B, which were previously impossible to run on the M4, can now be used without problems.

Initial token generation times show that the M6, with its 0.72-second success rate, clearly surpasses the M4’s 2.5-second result.

Image Rendering Times Significantly Reduced

A similar success story emerges in image rendering capacity. In tests conducted with the FLUX.1 model, the M6 ​​chip renders an image every 36 seconds, while the M4 processor completes the same task in over a minute and a half. This represents a 61.7% speed increase.
As you can see, this speed increase not only saves time but also increases power efficiency by reducing power consumption from 44 watts for the M4 to 38 watts.

However, it’s worth noting that this high performance comes with a price increase of approximately $300.

Do you think the high AI performance achieved by Apple’s M6 chip is worth the $300 price difference for professional users? Share your experiences working with local AI models and your thoughts on the new M6 in the comments section.