WWDC 2026: The Siri AI We Waited Four Years For Is Here, and the Engine Is Google's Gemini

At the WWDC 2026 keynote on June 8, Apple delivered the thing it has owed users for four years: a genuinely rebuilt Siri.

Reuters framed it bluntly: investors wanted to know whether AI can save Siri. From the promises at WWDC 2024 through the delays and the team reshuffles, “the new Siri” had become Apple’s most credibility-draining product commitment. The Guardian noted another milestone: this was Tim Cook’s final WWDC keynote.

Siri AI: Apple’s shell, Google’s engine

The new assistant is called Siri AI. Apple’s press release describes it as more conversational, context-aware, and able to carry out multi-step tasks across apps.

The real news is the engine. TechRadar’s live coverage recorded Apple publicly acknowledging its partnership with Google for the first time on a keynote stage: upcoming Apple Intelligence features are powered by Gemini models. Per MacRumors, a custom Gemini runs on Apple’s own Private Cloud Compute infrastructure, and multiple reports put the deal at about $1 billion a year.

The architecture choice deserves a second thought. Apple didn’t outsource Siri to the Gemini app. The model comes in, the data doesn’t go out — inference runs on servers Apple controls, which keeps the privacy story intact. A year ago everyone was guessing whether Apple would pick OpenAI or Google. Now we have the answer, and ChatGPT’s exclusive spot inside iOS is gone. Tom’s Guide notes OpenAI won’t be happy about it.

Beyond Siri

TechCrunch’s roundup lists the Apple Intelligence updates around the edges: tab management in Safari, one-tap password updating, and cross-app context awareness. On the system side there’s iOS 27 and macOS 27 Golden Gate, with adjustments to the Liquid Glass design language, plus the child safety features NPR highlighted — Apple spent part of the keynote shipping AI features and part of it criticizing, without naming names, how its competitors handle theirs.

The timeline: developer betas are out now, public beta in July, general release in the fall with the new OS versions.

Background: the January deal

In January 2026, Apple and Google announced a multi-year collaboration under which the next generation of Apple Foundation Models is based on Google’s Gemini models and cloud technology (joint statement; Reuters). The statement says Apple Intelligence continues to run on Apple devices and Apple’s Private Cloud Compute, under Apple’s privacy standards. In other words, Gemini technology underpins Apple’s models, but requests are processed on Apple’s own infrastructure rather than being handed to the Gemini app.

What developers get

Apple’s developer announcement adds several pieces:

• App Intents updates let apps connect to Siri AI capabilities such as personal context, app actions and on-screen awareness.
• The Foundation Models framework becomes a single Swift API for on-device models (now with image input), server models and custom skills. Developers in the App Store Small Business Program with fewer than 2 million total first-time downloads can use the next-generation Apple Foundation Models on Private Cloud Compute at no cloud API cost. Models from other providers, such as Claude or Gemini, can be plugged in through a new language model protocol.
• Core AI is a new framework for running custom models on device, optimised for Apple silicon.
• Xcode 27 brings coding agents from Anthropic, Google and OpenAI into the IDE, with planning, previews and the ability for agents to run tests and validate their own work.

Availability and requirements

According to Apple’s WWDC26 announcement:

• Timing. Developer testing began on the day of the keynote, a public beta was scheduled for the following month through the Apple Beta Software Program, and the features arrive as a free update in the fall with iOS 27, iPadOS 27, macOS 27, watchOS 27 and visionOS 27.
• Siri AI is labelled beta. It will be available as a beta later in the year for supported devices set to English, with more languages to follow.
• Supported devices. iPhone 15 Pro and Pro Max, iPhone 16 models or later, iPad mini (A17 Pro), iPads and Macs with M1 or later, MacBook Neo (A18 Pro), Apple Vision Pro, and Apple Watch Series 9 or later, Ultra 2 or later and SE 3 when paired with an Apple Intelligence-enabled iPhone.
• Regional limits. Siri AI is not initially available in the EU on iOS, iPadOS and watchOS (Mac and Vision Pro users in the EU can use it in a supported language). Siri AI and the other new Apple Intelligence features are not available in China while Apple works through regulatory requirements.
• Usage limits. Some features that rely on server models, including image generation, have daily usage limits; most iCloud+ plans raise them.

Our take

This WWDC reads best side by side with Google I/O 2026. Three weeks ago Google pushed Gemini into every entry point it owns. Now, through Apple’s own keynote, Gemini walked into the iPhone too. Gemini holds system-level positions on both Android and iOS, distribution no other lab has — not even Anthropic, which passed OpenAI in valuation last month.

For Apple, this is a pragmatic concession: building its own frontier model wasn’t going to happen on any useful timeline, so it bought the best engine available and kept the things it actually owns — hardware, privacy, distribution. What’s left is the experience question. Siri AI looked smooth in the demos, but Siri carries heavy baggage, and whether the shipping version can survive four years of accumulated expectations is a fall question. We’ll publish usage notes once the public beta opens in July.