Google limits Gemini Intelligence to recent flagship Android phones

Google is drawing a much tighter circle around its newest Android AI features than its broader Gemini branding suggests. The company said on May 12 that Gemini Intelligence will arrive first on “the latest Samsung Galaxy and Google Pixel phones” this summer, framing the package as something built for its “most advanced devices.” (blog.google) Google has pitched Gemini Intelligence as a set of proactive, on-device and cross-app features that can automate tasks, rewrite spoken thoughts into cleaner messages, summarize web pages and build custom widgets from natural-language prompts. (9to5google.com) But Google’s own developer and Android materials show those experiences depend on newer on-device AI plumbing, including AICore and Gemini Nano, rather than the lighter requirements for the standard Gemini mobile app. (blog.google) That distinction matters because the Gemini app itself runs on many Android devices with as little as 2GB of RAM and Android 9 or later, according to Google support pages. (blog.google) Gemini Intelligence, by contrast, is being positioned as a premium layer tied to newer hardware and software support commitments. ### Which phones is Google actually talking about? Google has not published a full public compatibility list on its main consumer pages, but it has named the launch cohort in narrower terms. The company said Gemini Intelligence features will roll out in waves starting with the latest Samsung Galaxy and Google Pixel phones this summer, and Google separately said it had spent months fine-tuning multi-step automation on the Galaxy S26 and Pixel 10. (blog.google) Samsung’s February announcement was more specific about one early device family. Google said task automation would start as a beta feature in the Gemini app on select devices such as the Galaxy S26 series, initially in the United States and South Korea, for selected food, grocery and rideshare apps. (support.google.com) ### What is the hardware bar? Google’s official Gemini Nano documentation says the model runs through Android’s AICore system service, which handles on-device inference and model updates. A separate Android Developers blog post said the Prompt API “currently performs best on the Pixel 10 device series,” which runs Gemini Nano v3, the latest version of the on-device model. (blog.google) Google’s consumer-facing Gemini Intelligence page does not spell out the full spec list in the visible text returned by search, but multiple reports citing Google’s product-page footnotes said the minimum requirements include a flagship-class chip, at least 12GB of RAM, Android AICore support and Gemini Nano v3 or higher. (blog.google) Those reports also said Google requires at least five Android OS upgrades and six years of security updates. Reuters could not independently verify the full footnote text from the page snapshot available through search results, but the cited requirements are consistent with Google’s emphasis on newer AICore-based on-device processing and Nano v3 support. (developer.android.com) ### Why does Gemini Nano v3 matter so much? Google’s October 2025 developer post tied Nano v3 to a newer architecture, saying the version is built on the same architecture as Gemma 3n. That matters because Gemini Intelligence is not just a cloud assistant skin; Google is presenting it as a local intelligence layer that can process prompts on-device through AICore with lower latency and stronger privacy controls. Google’s own materials say on-device generative AI avoids server calls, can work offline and keeps sensitive data on the device, but they also say inference speed depends on device hardware. (9to5google.com) That means the feature set is likely to track hardware capability more closely than older assistant rollouts that could rely more heavily on the cloud. ### How is this different from the Gemini app most Android users already have? Google support pages draw a clear line between the app and the new platform layer. The Gemini mobile app is available on many Android phones and tablets with 2GB of RAM or more, while Gemini Intelligence is being marketed separately as “the best of Gemini on our most advanced devices.” (developer.android.com) Google’s May 12 announcement also described capabilities that go beyond chat. The company said Gemini Intelligence can automate multi-step tasks across apps, summarize and compare content in Chrome, clean up voice-to-text output through a feature called Rambler and create widgets from plain-language prompts. (developer.android.com) ### When will more devices get it? Google said on May 12 that the phone rollout begins this summer on select Samsung and Google devices. The company also said Gemini Intelligence will expand later in 2026 to other Android form factors, including watches, cars, glasses and laptops, and its Android site says users can sign up to be notified as those releases approach. (support.google.com) (blog.google)