© ROOT-NATION.com - Use of content is permitted with a backlink.
As confirmed by Bloomberg and journalist Mark Gurman, iOS 27, iPadOS 27, and macOS 27 will feature a system called Extensions, which will allow users to choose between several third-party AI models, including Google Gemini and Anthropic Claude, to power Siri, Writing Tools, and Image Playground. The Extensions mechanism works through the App Store. AI service developers integrate support into their applications, and users can choose the desired model in Apple Intelligence and Siri settings. Requests within Apple functions are redirected to the selected provider.
This eliminates ChatGPT’s exclusive status as the only external AI model in Apple Intelligence, which has existed since iOS 18 and has already become the subject of an antitrust lawsuit by Elon Musk’s xAI due to the alleged restriction of competition.
At the same time, Gemini retains a separate privileged position as the base model for the updated Siri and key features of Apple Intelligence. According to reports, this integration is supported by a contract worth about $1 billion annually. This means that Extensions does not replace the existing architecture, but is overlaid on top of it. Users will also be able to assign different Siri voices to different AI models, which makes the source of the answer not only technically but also audibly identifiable.
Extensions is essentially a system for accessing generative AI capabilities from installed apps on demand via Apple Intelligence. As part of Siri, Writing Tools, Image Playground, and other features, users can send tasks to the chosen provider. For example, after installing the Claude app and activating Extensions support, requests can be made through Claude instead of the standard Apple model. This extends the approach that previously worked only as a backup for ChatGPT in iOS 18 to the entire Apple Intelligence ecosystem.
The distribution of functions is realized through the App Store. Providers add support for Extensions to their apps, Apple creates a special section of AI apps to promote them, and users customize the choice in system settings. Apple retains control over access criteria, display mechanism, and standard settings, which ensures openness within its own rules.
The separation of Gemini’s roles is important. At the system level, it is used as a basic model for Siri and Apple Intelligence, which works independently of user settings. Separately, through Extensions, Gemini can be made available as an optional choice alongside Claude and other models. Thus, the same AI provider functions in two different access levels.
A user who does not change the Extensions settings can still interact with Gemini through system integration. A user who chooses Claude intentionally changes the route of request processing. This creates a fundamental difference between automatic and selective use of models.
Siri’s voice differentiation function is worth special attention. Assigning different voices to different AI models makes the source of the answer obvious by ear. This demonstrates that models are not interchangeable and have their own characteristics, even if they work within the same Apple Intelligence system.
WWDC 2026 will take place from June 8 to 12, where Apple will officially present the Extensions architecture, including access conditions, privacy policy, and standard system settings.
Read also:
- Apple upgrades Wallet: Users will be able to create digital passes
- Secrets of watchOS 27: Apple will change the look of your watch