Apple said Monday it expanded Xcode and the Foundation Models framework at its 2026 Worldwide Developers Conference (WWDC), giving developers image input, custom skills and server-side model execution for apps built around Apple Intelligence. The developer documentation says the framework can now route work to on-device models, Private Cloud Compute or another server model provider when a task needs more reasoning capacity or context. The announcement pushes Apple's AI work deeper into developer tools, tying Foundation Models and App Intents to a broader Siri AI systemwide push.

Key Takeaways

AI-generated summary, reviewed by an editor. More on our AI guidelines.

Foundation Models gains images

Apple's updated documentation describes Foundation Models as a beta framework for iOS 26, iPadOS 26, Mac Catalyst 26, macOS 26, visionOS 26 and watchOS 27, with APIs for language understanding, structured output and tool calling. The page lists three new areas: server-side intelligence with Private Cloud Compute, dynamic sessions with instructions and profiles, and multimodal prompting for image analysis.

The same documentation says on-device models handle summarization, entity extraction, text and image understanding, refinement and game dialog. Developers can use Apple's @Generable macro to request Swift data structures and the Tool protocol to let a model call app code, such as a local database search or an app service request.

Xcode adds agent tools

MacRumors reported from the WWDC keynote that Xcode's coding assistant can now handle app localization and interact with simulated devices. Apple also said developers will be able to resize and work with app previews, with further technical sessions due during the conference.

Craig Federighi, Apple's senior vice president of software engineering, said Xcode is now the best place to build apps using agentic coding, according to MacRumors. MacRumors reported that Apple also announced a Core AI framework. MacDailyNews' live notes said Core AI lets developers use other models, while Xcode gains the ability to use a model and agent of the developer's choice.

Server models sit beside local AI

Apple's machine-learning research group said in a 2025 foundation-model update that its Apple Intelligence stack includes an approximately 3-billion-parameter on-device language model and a server model built for Private Cloud Compute. That report said the newer models support text and image inputs, have tool-use and reasoning improvements, and are designed to support 15 languages.

The June 2025 Apple Newsroom release said the Foundation Models framework gives developers no-cost AI inference, offline operation and native Swift support. It also said Xcode 26 let developers connect large language models into coding workflows through built-in ChatGPT support, API keys from other providers or local models on Apple silicon Macs.

Siri AI sets the developer target

CNBC reported Monday that Apple created a second version of its Apple Foundation Models that can understand speech, text and images. CNET and TechCrunch reported that Apple worked with Google and Gemini technology on the next generation of Apple Intelligence, matching Apple's earlier privacy-boundary framing.

The hardware floor will matter for developers. 9to5Mac reported that Apple's most powerful on-device model requires an iPhone 17 Pro or iPhone Air, an M4 or later iPad with 12GB memory, or an M3 or later Mac with 12GB memory. CNBC and 9to5Mac also reported that Siri AI will not be available in the EU or China at launch because of regulatory issues. Apple says broader Foundation Models access still depends on users enabling Apple Intelligence on supported devices.

Frequently Asked Questions

What did Apple change in Foundation Models at WWDC 2026?

Apple added image analysis, dynamic sessions, server-side intelligence through Private Cloud Compute and APIs for structured output and tool calling.

How does Xcode fit into Apple's AI developer update?

Apple said Xcode's coding assistant can handle localization, interact with simulated devices and work with a developer's chosen model or agent.

What is Private Cloud Compute in this context?

Private Cloud Compute is Apple's server-side path for model requests that need more reasoning capacity or context than on-device models can provide.

Which devices can run Apple's strongest on-device model?

9to5Mac reported that the top model requires an iPhone 17 Pro or iPhone Air, or newer iPads and Macs with 12GB memory.

Why does this matter for third-party apps?

The new APIs give developers Apple-controlled ways to add model calls, image understanding, structured output and app actions.

AI-generated summary, reviewed by an editor. More on our AI guidelines.

AI News

San Francisco

Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: editor@implicator.ai