Apple announces the Foundation Models framework, which for the first time gives developers direct access to its on-device AI model powering Apple Intelligence
Apple at WWDC today announced Foundation Models Framework, a new API allowing third-party developers to leverage the large language models …
MacRumorsTim Hardwick
Context & Ripple Effects
Apple’s move turns Apple Intelligence from a system-level feature set into a developer-facing capability: third parties can build against the on-device model rather than only consume Apple’s finished experiences.
Developers gain direct API access to the on-device model behind Apple Intelligence, enabling AI features that can run within Apple’s platform without requiring a separate model integration.
Apple becomes the provider of a new application-layer AI interface, making its model capabilities a product surface for its developer ecosystem.
Second-order effects
App teams can weigh native on-device integration against external AI services, with Apple’s framework potentially reducing integration friction for use cases that fit its model and platform constraints.
The framework raises the value of Apple-specific AI tooling; the related Core AI and agentic-coding additions reinforce a more integrated developer workflow around Apple’s stack.
Third-order effects
If Apple continues broadening model access across device and cloud, AI differentiation in its ecosystem may shift from standalone assistants toward applications built on a common platform model layer.
This is also a model-governance choice: platform owners increasingly define which model capabilities developers can access and under what technical boundaries.
The trend: Consumer-platform companies are turning proprietary foundation models into developer infrastructure, with on-device access becoming a key way to extend AI features across their app ecosystems.
Apple has announced the Foundation Models framework which will let developers leverage its on-device AI models. — The pitch to developers is that they'll save cloud API costs. — The counterpoint is that if you have an Android app then you end up with vastly differing AI in yo…
Apple exposing a developer API for their on device AI models is a major step in the right direction for them and the open model ecosystem. This is going to open up new feedback loops on models and a major platform for AI applications. [image]
Excited to see how it goes! They should release the weights too to accelerate its adoption (even if it barely makes sense on other devices). Hopefully this makes it out of Beta unlike some other dev features (I'm a noob about wwdc). https://developer.apple.com/ ...
Apple doesn't report benchmarks for their AIs, reporting on an ill-documented head-to-head evaluation But even by their standards, Apple's latest on device models are mostly worse than the open Gemma 3-4B from Google or Qwen 3-4B And their server LLM is similar to Llama 4 Scout […
The most important WWDC update today was the announcement of the Foundation Model framework, which makes it easier for developers to integrate Apple Intelligence on device. It matters because, just like when the App Store launched in 2008, it's up to developers to build the apps…
.@benthompson Apple listened to you opening up on-device models!! “Experiences that are smart, available when you're offline, and protect your privacy.” ...and offers developers the value that Apple uniquely can with a billion devices at the edge with free compute...
The big focus that they're emphasizing is the Foundation Models framework. It seems there is a way to help things integrate between apps and media, even offline and privately—lots of questions. [image]
I debated this with @_chuckyc recently, the idea of free AI use sounds great, but for mosts tasks, you want the best model possible, and that's not an Apple model on device.
New@axios: Apple opens its AI models up to developers, announces new Liquid Glass design interface even as critics worry the company is falling behind [image]