Our editorial team is comprised of skilled technology experts and developers. To ensure that our research is easy to understand in simple and plain English, we may use AI-assisted tools for grammatical refinement and structural smoothness. However, every technical insight, test, and experience displayed has been fully completed and verified by our human team. All content remains the original property of Droid Expose. See more in our Privacy Policy.
In April, we covered the moment Google Cloud CEO Thomas Kurian confirmed on stage that Gemini would power the next generation of Apple Intelligence. At the time, the practical details were thin. We knew Gemini was the engine. We did not know what that meant for the millions of developers who build the apps people actually use on their iPhones every day.
On June 9, one day after Apple’s WWDC keynote, Google answered that question directly. Gemini models are now available inside Apple’s own Foundation Models framework, and Gemini is built into Xcode itself, the software every iOS developer uses to write code. This is not a consumer feature. It is the plumbing underneath the consumer features, and it tells you something the keynote stage did not: how seriously both companies are treating this as infrastructure, not a one-time integration.
Table of Contents
What Actually Got Announced
According to Google’s own developer blog, Apple opened its Foundation Models framework to third-party cloud model providers starting with iOS 27, macOS 27, iPadOS 27, visionOS 27, and watchOS 27. Model providers can now implement something called the LanguageModel protocol, which gives every model, whether it is Apple’s own on-device model or a cloud model like Gemini, a shared, common interface.
In plain language: a developer building an app can write one piece of code and switch between Apple’s built-in AI model and Google’s Gemini model without rewriting how their app talks to either one. Google made this possible by hooking Gemini into the framework through the Firebase Apple SDK, specifically through something called Firebase AI Logic, which lets developers plug Gemini directly into their iOS, macOS, iPadOS, and visionOS apps without standing up their own backend server.
Google published a short code sample showing how small the actual change is for a developer who already uses Apple’s framework. Swapping from Apple’s on-device model to a cloud-hosted Gemini model is, in their words, a small code change, essentially swapping the model instance while everything else about how the app calls it stays the same.
Why This Matters More Than It Sounds
This is the part that connects back to what we wrote about Apple’s Extensions feature for Siri. We reported in May that Apple was building a system letting users choose which AI model powers features like Siri, Writing Tools, and Image Playground, with Google and Anthropic both confirmed as testing partners.
What Google just announced is the developer-facing half of that same idea. Extensions is the user-facing switch. The Foundation Models framework opening to Gemini is what makes that switch technically possible at the app level, not just inside Siri itself. A developer building a travel app, a writing app, or a customer service tool can now choose to run their AI features on Gemini specifically, inside an Apple-native development environment, using Apple’s own framework conventions.
That is a meaningfully different relationship than Apple simply licensing Gemini to power Siri in the background. This is Apple inviting Google’s models to sit alongside its own as a first-class option for every developer on its platform.
Gemini Is Now Inside Xcode Itself
The second half of Google’s announcement is about Xcode, Apple’s development tool that every iOS and Mac app gets built in.
Google says it worked directly with Apple to integrate Gemini into Xcode so developers can perform complex, multi-step coding tasks without switching to a separate tool or browser tab. Once a developer sets this up through the Intelligence settings panel inside Xcode, Gemini can review code, help fix bugs, and assist in building new features, functioning as what Google calls an agentic experience inside the editor itself.
This is worth pausing on because of how unusual it is. Apple has historically kept its development tools tightly closed. Having a competitor’s AI model embedded as an option directly inside Xcode, with Apple’s cooperation, is a notable shift in how Apple is willing to work with outside AI providers on its own home turf.
For authentication, Google is offering two paths. Individual developers can get a self-serve Gemini API key from Google AI Studio, which includes both a free tier to get started and a paid tier for more advanced models and higher volume use. Enterprise development teams can instead go through the Gemini Enterprise Agent Platform, which gives them access to their organization’s dedicated corporate quotas and data privacy settings.
How This Lines Up With What Apple Said at WWDC
Apple’s own WWDC announcement, published the same week, confirmed that the next generation of Apple Foundation Models was custom-built in collaboration with Google and its Gemini models for deeply integrated Apple Intelligence experiences. Craig Federighi’s framing on stage, that Siri should feel like an integral but conversational tool rather than a separate chatbot, lines up with what Google’s developer announcement actually enables. If Gemini sits behind the same interface as Apple’s own model, app behavior feels consistent regardless of which model is doing the underlying work. That consistency is what makes a deeply embedded AI assistant feel native rather than bolted on.
It is also worth noting that Apple’s privacy architecture does not change because Gemini is now an option. Apple’s announcement reiterated that its new Apple Foundation Models run on device and on servers using Private Cloud Compute, with personal data not stored or made accessible to Apple or anyone else when Private Cloud Compute is handling a request. Whether that same privacy standard extends cleanly to requests routed through Gemini specifically is not detailed in either company’s announcement, and is a genuine open question developers and privacy-conscious users should watch for as real apps start shipping with this integration.
What This Means for Regular iPhone Users
None of this is a feature you will personally tap on. But it shapes what is coming.
Every developer who builds a writing tool, a research assistant, a customer support chatbot, or any AI-powered feature inside an iPhone app now has an easy, Apple-sanctioned path to use Gemini specifically, rather than building a custom integration from scratch or defaulting only to Apple’s own model. That likely means more apps shipping with genuinely capable AI features sooner, because the technical barrier to using a frontier model dropped significantly this week.
It also reinforces something we said back in April: this partnership is not a temporary arrangement to cover a gap until Apple’s own models catch up. Building Gemini into the actual developer tooling, not just the consumer-facing Siri experience, is the kind of integration companies do when they expect a relationship to last for years, not one product cycle.