Apple to let Google and Anthropic models plug into Apple Intelligence in iOS 27

Apple appears to be moving Apple Intelligence toward a multi-provider model in iOS 27, according to reports published in early May and this week. Bloomberg reported on May 5 that Apple plans to let users choose from multiple outside AI services across its software, including for generating and editing text and images. 9to5Mac and other Apple-focused outlets said the shift would fit a broader iOS 27 push expected to be presented at WWDC on June 8. That would be a notable change from Apple’s current setup. Apple’s own support page says Apple Intelligence today can use a ChatGPT extension for Siri and Writing Tools on iPhone. A separate report from Quasa said Apple is testing Google’s Gemini and Anthropic’s Claude for a wider set of Apple Intelligence functions, while keeping Apple’s own on-device models for lighter, privacy-focused tasks. (bloomberg.com) ### Which Apple features are reported to be opening up? Bloomberg’s May 5 report said the change would cover features across Apple’s software, with users able to select third-party AI models for tasks such as text and image generation. 9to5Mac said internal test versions describe “Extensions” that let installed apps provide generative AI capabilities on demand through Siri, Writing Tools and Image Playground. (support.apple.com) MacRumors reported on May 5 that Apple had signed a deal with Google to use a Gemini-based model for some Apple Intelligence and Siri features in iOS 27, while still allowing users to choose another service. That report also said OpenAI would no longer be the only outside option exposed through Apple’s AI system. (bloomberg.com) ### Does this replace Apple’s own models? Quasa reported on May 21 that Apple’s own models would continue handling lightweight or privacy-sensitive requests, with outside providers used for more complex tasks. That matches Apple’s current public positioning around Apple Intelligence, which combines on-device processing with optional handoff to ChatGPT for deeper responses. (macrumors.com) The reported architecture matters because Apple would be adding choice without fully outsourcing the assistant layer. None of the cited reports said Apple planned to stop using its own models; they described a system that routes some requests to external providers when needed or when selected by the user. That is an inference from the reports and Apple’s existing support documentation. (quasa.io) ### Why do Google and Anthropic matter here? Google and Anthropic are the two outside model providers named most consistently in recent reports. Quasa said Apple is conducting closed-door testing with Gemini and Claude. MacRumors and 9to5Mac, both citing Bloomberg’s reporting, said Apple wants Siri and related AI features to work with rival chatbot services beyond ChatGPT. (quasa.io) That would give Apple more than one large-model partner at a time. It would also reduce dependence on a single outside provider for Apple Intelligence features, according to the reports. ### What has Apple confirmed publicly? Apple has publicly confirmed only the current ChatGPT integration, not the reported iOS 27 expansion to Google and Anthropic. (quasa.io) Apple’s support documentation says users can enable the ChatGPT extension inside Apple Intelligence on iPhone. Apple has not, in the sources reviewed here, published an equivalent support page or announcement for Gemini or Claude integration in Apple Intelligence. (bloomberg.com) WWDC 2026 is scheduled for June 8 through June 12, according to Apple-focused coverage surfaced in search results, and multiple reports say iOS 27, iPadOS 27 and macOS 27 are expected there. If Apple follows that timetable, the next concrete test for these reports will be the WWDC keynote and related developer materials. (apfelpatient.de) (support.apple.com)