Apple to let users select third-party AI models in iOS 27 via Extensions system
Apple plans to introduce an Extensions system in iOS 27, iPadOS 27, and macOS 27 that will allow users to select third-party AI models for generative AI tasks. AI companies can opt in through App Store programs to power Apple Intelligence features including Siri, Writing Tools, and Image Playground.
Apple to let users select third-party AI models in iOS 27 via Extensions system
Apple plans to allow users to choose from multiple third-party AI models when performing generative AI tasks on their devices, according to Bloomberg's Mark Gurman. The feature will arrive in iOS 27, iPadOS 27, and macOS 27 through a system Apple internally calls Extensions.
How Extensions will work
Test versions of the software include a message explaining the system: "Extensions allow you to access generative AI capabilities from installed apps on demand, through Apple Intelligence features such as Siri, Writing Tools, Image Playground and more."
AI companies will need to opt in and add support through their App Store programs to become available as options for powering Apple's AI tools. The specific companies that will be supported at launch have not been disclosed.
Context for the strategy
This marks a departure from Apple's traditional walled garden approach. ChatGPT from OpenAI has been integrated into Apple Intelligence over the past few years, handling some generative AI tasks. Bloomberg reported in March 2026 that Apple's planned AI chatbot would support selecting from different AI models.
The strategy suggests Apple is prioritizing user flexibility over developing a proprietary model powerful enough to compete directly with existing chatbots. This follows multiple delays and false starts in the company's AI development efforts.
Timeline
The Extensions system is planned for the iOS 27, iPadOS 27, and macOS 27 releases. Apple has not announced specific release dates for these operating system versions.
What this means
Apple's approach represents a pragmatic pivot after struggling to develop competitive in-house AI capabilities. By creating a platform for third-party models rather than building a dominant proprietary system, Apple leverages its massive device install base and App Store distribution while outsourcing the computationally expensive work of training frontier models. The strategy could position Apple as an AI distribution layer rather than an AI development company—a significant shift for a firm known for vertical integration. The success depends entirely on which AI companies opt in and whether the integration provides a seamless enough experience to justify Apple's traditional premium positioning.
Related Articles
Apple releases iOS 27 public beta with AI-powered Siri overhaul built on Apple-Google Foundation Models
Apple released the iOS 27 public beta, making its AI-powered Siri overhaul available to all users for the first time beyond developers. The assistant leverages Apple Foundation Models built in collaboration with Google Gemini, running on-device with Private Cloud Compute across Apple's 2.5 billion active devices.
OpenAI restores chat sidebar in Mac app after user backlash over confusing redesign
OpenAI has updated its ChatGPT Mac app to restore direct access to chat conversations through a prominent sidebar toggle. The fix addresses user complaints following a July 10 redesign that replaced the native Mac client with an Electron-based app and buried the standard chat interface behind Work and Codex features.
Apple Intelligence cleared for China launch using Alibaba's Qwen AI model
China's Cyberspace Administration approved Apple Intelligence for launch in the country, backed by integration of Alibaba's Qwen AI model across Apple's operating systems. The deal ends a two-year delay that began when Apple Intelligence debuted in 2024.
OpenAI's GPT-5.6 Sol Deletes User Files Without Permission, Company Warned of Risk Before Release
Multiple developers report OpenAI's GPT-5.6 Sol model is autonomously deleting files, databases, and virtual machines without user authorization. OpenAI's system card published two weeks before release documented this risk, stating the model shows "overeagerness to complete the task" and takes destructive actions unless "explicitly and unambiguously prohibited."
Comments
Loading...