Google Launches Gemini Nano: Local AI Now Accessible for App Developers

Mon 19th May, 2025

Google has introduced Gemini Nano, a local artificial intelligence (AI) solution that is now available for Android app developers. With the launch of a new API, this AI can assist in various tasks, including text summarization and paraphrasing.

Through the ML Kit GenAI interface, third-party applications can leverage the capabilities of Gemini Nano to enhance user experience by providing features like writing aids and other functionalities directly on Android devices. However, initial support appears limited to devices with higher processing power, leaving some lower-end smartphones without access to these features.

The new ML Kit GenAI interface, while not officially announced by Google, has been detailed in developer documentation that was discovered by Android Authority. This release is anticipated to be officially unveiled during the Google I/O 2025 event scheduled for May 20.

According to the documentation, developers will have the ability to utilize Gemini Nano within their applications to assist users with basic tasks. These AI capabilities are powered by AICore, a system service in Android that enables on-device execution of generative AI models.

The features provided by Gemini Nano are comparable to those offered by Apple in their system applications, such as Mail and Notes, which incorporate Apple Intelligence with iOS 18.4 and macOS. The functionalities include text summarization, correction suggestions, and image descriptions, along with the ability to rephrase short messages in various tones and styles. Samsung also offers similar capabilities on select smartphones through Galaxy AI.

Initially, not all features of Gemini Nano will be available in every language. For instance, the summarization feature, which generates up to three key points from a text or chat history, will only be offered in English, Japanese, and Korean. Meanwhile, the proofreading and rewriting functions will support seven languages, including German, while image description capabilities will initially be limited to English.

One significant advantage of utilizing local AI is the enhanced privacy it offers, as personal data does not need to be transmitted to Google's cloud servers. This approach not only protects user privacy but also improves the speed of output since data does not have to be sent back and forth. Furthermore, these AI functionalities can be used offline, making them accessible even in flight mode or in areas without cellular service. Developers also benefit from the absence of additional server costs associated with each API call, as stated by Google.

However, the new on-device AI features require sufficient processing power and memory, which may not be met by entry-level or mid-range smartphones. Supported devices include Google's latest Pixel 9 series, high-end models from Samsung's Galaxy S25 series, as well as premium devices from Xiaomi's 15 series, and top-tier products from Oppo, OnePlus, Honor, and Motorola. Notably, Google's own mid-range Pixel 9a is not on the list of supported smartphones. Google has indicated that support for additional devices will be expanded in the future.

It appears that the new ML Kit GenAI interface will either replace or augment Google's previously used experimental AI Edge SDK, which has primarily been utilized with the Pixel 9 series.


More Quick Read Articles »