Foundation Models framework: Apple releases new Swift API
Apple has introduced the Foundation Models framework, a native Swift API designed to integrate any large language model into apps.

Apple has released the Foundation Models framework, providing developers with a native Swift API to integrate any large language model into their applications. Whether generating quick summaries or creating dynamic AI agents, the framework offers comprehensive capabilities for modern app development.
- Native Swift API for integrating language models into Apple platform applications.
- Direct access to on-device Apple Intelligence models and server-hosted models like Claude and Gemini.
- Built-in multimodal capabilities, including OCR, barcode scanning, and image understanding.
- Advanced features like Dynamic Profiles, Evaluations framework, and a Python SDK.
On-Device and Cloud Model Integration
The new framework grants direct access to the core on-device foundation model of Apple Intelligence, enabling smart, private experiences that function without internet connectivity. When more complex workflows are required, developers can seamlessly extend the framework to server models such as Claude and Gemini using the exact same Swift API.
Qualifying developers in the App Store Small Business Program with fewer than two million downloads can access Apple Foundation Models running on Private Cloud Compute with no cloud API cost. This ensures frontier-level intelligence alongside strict data privacy protections.
Read alsoMac admins are shipping AI-written code fast, but fixing errorsAdvanced Features and Developer Tools
Multimodal prompt support allows apps to reason about visual content by passing images alongside text, while the integrated Vision framework provides precise text extraction and barcode scanning. Additionally, the fm command line tool in macOS 27 helps streamline scripting and prototyping.
Developers also benefit from Dynamic Profiles for flexible session management, the Evaluations framework for rigorous prompt testing, a Python SDK for broader tool compatibility, and experimental utilities designed to accelerate AI pattern adoption.
Impact on App Development
The introduction of the Foundation Models framework simplifies AI adoption across Apple platforms, offering a unified protocol for both local and cloud-based language models. This empowers developers to build unique, secure, and context-aware intelligence features efficiently.
Source: Hacker News: Front Page



