WWDC26: Explore numerical computing in Swift with MLX | Apple
Apple's open-source MLX Swift framework brings NumPy-like array computing with automatic GPU execution and differentiation to numerical computing.
“If your primary goal is writing mathematical code with an eye for performance, MLX Swift is a great solution.”
At WWDC26, Apple engineer David Koski introduced MLX Swift, an open-source (MIT) framework for numerical computing that uses n-dimensional arrays like NumPy, with lazy evaluation enabling automatic GPU execution and differentiation. While foundational to on-device ML training on Apple platforms, this session focuses on developer tooling and numerical examples rather than a major AI industry shift.