How Apple's on-device AI works, and what it changes for your users

In 2026, Apple introduced its third-generation foundation models, known as Apple Foundation Models (AFM 3), which significantly enhance on-device AI capabilities by allowing iPhones to process tasks such as reading, summarizing, and ranking notifications, emails, and messages locally. The models operate on a mixture-of-experts design that activates only a small portion of the 20-billion-parameter model per request, thus enabling efficient performance on iPhones with A17 Pro chips or newer. This advancement means any app can leverage these models for free and offline, transforming how users interact with their phones by prioritizing concise, clear communication over vague or promotional content. The models ensure privacy by keeping data on the device and emphasize instant, cost-free processing, which allows users to rely on quick summaries rather than engaging with full content, fundamentally shifting the dynamics of user engagement and content delivery. Additionally, Apple's collaboration with Google on server models using Google's Gemini technology marks an expansion of its AI capabilities into more complex tasks handled in data centers, while maintaining privacy and security commitments.

No tracked trend matches for this post yet.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.