Apple to Enhance Siri with Google's Gemini Model, Launching in 2025

Apple is reportedly developing a new feature codenamed "World Knowledge Answer" that will be integrated into Siri, offering capabilities similar to Google Search's AI overview and AI search functions like Perplexity and ChatGPT. This feature is expected to run on a custom version of Google's Gemini model and will be deployed on Apple's own cloud servers. The new feature will provide users with structured, easy-to-read summaries based on web data, supporting a mix of text, images, and videos. It may also be integrated into the Safari browser and the "Focus" search feature accessible by swiping down from the home screen.

Apple has reportedly reached a formal agreement with Google to test the Gemini model, which could enhance Siri's capabilities. However, Apple will continue to use its own Apple Foundation Model for local data searches. This collaboration comes after a U.S. judge ruled that Apple could maintain its default search engine agreement with Google, which generates approximately 200 billion dollars in annual revenue for Apple. The high-level vice president of Apple Software and Services had previously noted a decline in Google search requests from Apple devices, attributing this to the rise of AI search and hinting at Apple's proactive measures to avoid being marginalized.

Earlier, Anthropic's Claude model was considered a top contender for Apple's Siri AI collaboration, but its high annual cost of 1.5 billion dollars made it less favorable. Apple is still considering potential partners for the planning component of Siri, with options including Google, Anthropic, or an in-house model. Notably, Apple had previously discussed integrating Google's Gemini chatbot into Siri but ultimately chose to collaborate with OpenAI, integrating ChatGPT into the iOS system. However, this feature has not yet been officially released to users.

Apple plans to launch the upgraded Siri as early as March 2025 with the release of iOS 26.4. This means that even if the upcoming iPhone 17 series, set to be unveiled next week, does not include the new AI features, users may have to wait until at least six months later to experience the fully revamped Siri.

Apple has been grappling with the decision of whether to collaborate with third-party partners or develop AI capabilities in-house. This dilemma is reminiscent of the challenges faced by Apple in the 1990s with the Copland operating system project. At that time, the company's outdated system prompted the then-CEO to seek external solutions, with rumors circulating about a potential acquisition of BeOS. Ultimately, Apple acquired NeXT, bringing back Steve Jobs to the company. Today, history seems to be repeating itself as Apple evaluates potential acquisitions, including Perplexity and the French startup Mistral, to bolster its AI capabilities. However, after thoroughly examining Perplexity's technology, Apple decided against the acquisition but continues to explore other transaction opportunities to enhance its talent pool.

Apple's AI team is currently experiencing unprecedented talent loss. The company's foundation model team leader and distinguished engineer was reportedly poached by Meta with an offer exceeding 2 billion dollars, leading to a chain reaction of departures. In just two months, over ten researchers have left, joining competitors such as Meta, OpenAI, and Anthropic. This talent drain underscores the intense competition in the AI sector and the challenges Apple faces in retaining key personnel.