TL;DR
- Apple is giving third-party apps direct, offline access to its on-device AI model.
- Developers can use Apple’s Foundation Models framework with just a few lines of code.
- Testing starts today, with full rollout this fall on supported devices.
Google’s been trying to weave AI into Android apps for years. Apple Intelligence may have received some criticism in the past for being a little slow to the party, but at today’s WWDC 2025, Apple just handed app developers the keys to its main LLM, with offline and instant access.
As part of a wave of updates to its Apple Intelligence platform, the company announced that developers can now tap directly into the on-device foundation model that powers these features. This means any iOS app — not just Apple’s own — can add generative AI capabilities without an internet connection.
Apple pitched examples like a quiz app that auto-generates questions from notes.
According to Apple’s press release, the Foundation Models framework lets developers access Apple’s large language model using as little as three lines of Swift code. The model supports guided generation, tool calling, and natural language processing, all locally and free of charge. Apple pitched examples like a quiz app that auto-generates questions from notes, or a hiking app that lets you search trails in plain language, even if you’re completely offline.
Because everything happens on-device, Apple emphasizes that it “protects privacy by design.” Google may have been quicker in baking AI into the OS, but Apple is making it available to every app in a more seamless manner.
The new developer tools and Apple Intelligence features are available for testing now via the Apple Developer Program. A public beta will arrive next month, and full access will roll out this fall to users with supported devices.
By
