The magic of machine learning The machine learning technologies in the M1 chip open up a world of possibilities for Mac apps. Machine learning lets apps build and apply models based on massive amounts of data — to do things like identify friends and family in photos and videos, interpret natural language for dictation, and even analyze audio to recognize laughter, applause, and more. And that data is processed on device to protect your privacy — all at lightning speed.

Neural Engine The dedicated 16‑core Neural Engine in the M1 chip can execute up to a staggering 11 trillion operations per second, powering workflows you couldn’t imagine before — like enabling the djay Pro AI app to isolate instrumentals and vocals of any song in real time.

Optimized for machine learning The entire M1 chip is engineered for machine learning. The CPU, GPU, and Neural Engine are each designed for different types of machine learning workloads — and the ML performance controller distributes ML tasks to the right places to maximize performance.

ML accelerators The two dedicated ML accelerators are built for fast matrix multiplication, executing up to one trillion operations per second — perfect for low-latency ML tasks that don’t require the massive power of the Neural Engine.