• btc = $76 924.00 -1 227.04 (-1.57 %)

  • eth = $2 119.60 -67.66 (-3.09 %)

  • ton = $1.93 0.02 (1.26 %)

  • btc = $76 924.00 -1 227.04 (-1.57 %)

  • eth = $2 119.60 -67.66 (-3.09 %)

  • ton = $1.93 0.02 (1.26 %)

18 May, 2026
2 min time to read

Google has introduced strict hardware requirements for Gemini Intelligence, its new generation of on-device AI features for Android, leaving most flagship smartphones released before 2026 outside the supported list.

To run the new tools, devices need a flagship-grade chip, at least 12 GB of RAM, and support for Gemini Nano v3. The last requirement effectively rules out nearly every Android phone released before this year.

The company unveiled Gemini Intelligence earlier this week and outlined the device requirements in a footnote on android.com. The criteria turned out to be more demanding than even owners of recent flagships had expected. The cutoff excludes the Pixel 9, Galaxy Z Fold7, Xiaomi 17, and OnePlus 13, effectively the entire premium Android lineup released before 2026, with only two exceptions.

What Gemini Intelligence is

Gemini Intelligence is a set of AI features that runs locally on the device, without relying on cloud processing. The main capability is the autonomous execution of multi-step tasks in the background, with the smartphone able to search for information on its own, process it, and interact with apps and websites on the user's behalf. The Gboard keyboard introduces a new tool called Rambler, which recognizes speech while understanding filler words and language switching within a single sentence. Another announced feature, Create my Widget, generates home screen widgets on demand.

For these features to run locally, the device needs the Gemini Nano v3 model. The new version is larger than its predecessor and shares its architecture with Gemma 3n, the open-weight model Google released earlier this year. The model's size and complexity are the main reason for the new hardware requirements.

Which devices qualify

Google's official requirements include a flagship-class processor, at least 12 GB of RAM, and support for the AICore system service along with Gemini Nano v3 or newer. Devices must also be guaranteed five years of Android updates and six years of quarterly security patches, and meet specific media quality criteria such as spatial audio, low-light photography, HDR, and annual GPU driver updates.

The real bottleneck, however, is Gemini Nano v3. According to reports, the list of compatible devices on Google's developer page consists almost entirely of 2026 smartphones. Among older models, only the Pixel 10 and the Oppo Find X9 series qualify. Every other device still runs on Nano v2 and falls short of the new requirements, even when chip and RAM specifications are well within range.

The first devices to receive Gemini Intelligence will be the foldable Samsung Galaxy Z Fold8 and Z Flip8, reportedly scheduled to launch in July. The Galaxy S26 and Pixel 10 will follow. All other Android users, including owners of the Pixel 9 and Galaxy Z Fold7, will remain on the current set of features.

An awkward question around Pixel 11

The most uncomfortable scenario for Google involves its own upcoming flagship. According to recent leaks, the base version of the Pixel 11 will ship with only 8 GB of RAM, which is 4 GB below Google's own stated minimum. That leaves three possibilities: the leak is inaccurate, Google plans to grant an exception for its own device, or the entry-level Pixel 11 will be unable to run the company's flagship AI feature at launch.

Google hasn't completely ruled out support for older devices. The current documentation describes which phones support Nano v3 today, but doesn't explicitly state that older devices can't receive it through a future update. That leaves room for select 2025 flagships to eventually gain access to Gemini Intelligence, though Google hasn't made any promises so far.