Apple Quietly Built a New AI Stack and It Runs on Your Device

Apple introduces two foundation language models behind Apple Intelligence: a ~3B on-device model optimized for Apple silicon and a scalable server model using a new Parallel-Track Mixture-of-Experts architecture. Together they deliver multimodal, multilingual AI with strong privacy via Private Cloud Compute, plus a Swift-first framework that lets developers integrate AI features in just a few lines of code.