Apple Opens Foundation Models to Any LLM Provider at WWDC

Apple's Foundation Models framework, introduced with iOS 26, gave developers free access to the on-device model powering Apple Intelligence. At WWDC 2026, Apple dramatically expanded the framework's capabilities, positioning it as the default AI layer for the iOS ecosystem.

The headline feature is the new LanguageModel protocol, which abstracts away the underlying model provider. Developers write to a single Swift API, and can route requests to Apple's on-device model, Private Cloud Compute (now free for smaller developers), or third-party providers including Claude, Gemini, and open-source models. Switching providers requires no code changes, just a configuration update. This effectively makes iOS the first major platform where AI model selection is a deployment-time decision rather than a build-time commitment.

Additional updates include image input support for multimodal queries, server-side model integration, context management APIs for maintaining conversation state across sessions, and tool-calling capabilities that enable agentic workflows natively within the framework. Apple also brought an LLM provider adapter API, allowing any model vendor to plug into the Foundation Models ecosystem.

For iOS developers, this makes Apple's ecosystem the lowest-friction deployment target for AI-powered features. The on-device model remains free, private, and offline-capable. For MENA-based mobile developers, this matters because it removes the API key and billing overhead that previously made on-device AI features economically challenging at scale, especially for Arabic-language apps serving price-sensitive markets.