Core ML
Run machine learning models on device, across the CPU, GPU and Neural Engine.
Overview
Core ML runs trained models right on the device, so predictions are fast, work offline and keep data private. Drop a model into Xcode and it generates a Swift class you can call like any other.
Convert models from PyTorch and other frameworks with Core ML Tools, and compress them with quantization and palettization to fit on device.
On device
Fast, private predictions that work offline.
Xcode integration
Every model gets a generated Swift interface.
Core ML Tools
Convert models from PyTorch and other frameworks.
Compression
Quantization and palettization to shrink models.
History
How it grew up.
Introduced
Arrives with iOS 11 and Xcode 9.
Create ML
Train your own models on the Mac.
On-device training
Core ML 3 lets apps personalize models on device.
Stateful models
MLTensor and stateful models for generative AI.