Generative artificial intelligence developments drive a deep transformation of mobile technology, which is currently approaching a pivotal shift. Remote servers provide the computational power for today’s advanced AI systems, but Google plans to shift these capabilities into our smartphones. The upcoming Google I/O event has generated significant excitement as it promises to reveal a new collection of developer APIs that will leverage Google’s Gemini Nano model to deliver advanced AI processing directly on devices. This strategic initiative demonstrates Google’s dedication to delivering advanced AI features to consumers directly through devices while enhancing privacy and system performance by reducing cloud infrastructure dependency.

Developer documentation released by Google has provided an informative look at upcoming AI upgrades. Android Authority investigative reports reveal that the upcoming ML Kit SDK update will allow developers to build generative AI features on devices through Gemini Nano-powered APIs. This new framework builds on Google’s AI Core, which serves a similar purpose as the experimental Edge AI SDK but stands out for its improved seamlessness and user-focused design. This solution connects developers to an existing model while delivering a precise set of features to streamline their work, which makes advanced AI tools available to more mobile app developers.

The detailed documentation provided by Google thoroughly explains how the latest ML Kit GenAI APIs enable applications to perform essential tasks directly on-device and remove the requirement for cloud processing of sensitive user information. These capabilities include:

• Text Summarization: The feature enables users to transform extensive text documents into summaries that are simple to understand.
• Proofreading: The proofreading function works by recognizing grammatical errors and typographical mistakes and then proposing appropriate corrections.
• Rewriting: The rewriting tool provides different phrasings and stylistic improvements to improve written content.
• Image Description: The system generates written descriptions of images automatically while maintaining accurate content representation.

The fundamental limitations in mobile device processing power require specific restrictions for the Gemini Nano version that operates on devices. The text summaries will be restricted to three bullet points, while the image description feature will initially be available only in English. The performance of AI-generated content can differ based on which version of Gemini Nano a specific smartphone uses. Gemini Nano XS has a relatively compact size of 100MB, but Gemini Nano XXS, which operates in devices like the Pixel 9a, is even more streamlined and occupies only 25MB while supporting text-based processing with a smaller context window.

Implications for the Android Ecosystem

The strategic shift by Google has major consequences for Android, as the ML Kit SDK supports devices beyond Google’s exclusive Pixel lineup. Gemini Nano already plays a central role in Pixel smartphones, and prominent Android manufacturers such as OnePlus (13 series) and Samsung (Galaxy S25), alongside Xiaomi (15 series), are reportedly updating their upcoming devices to support this AI model directly. As support for Google’s local AI model grows across Android smartphones, developers will reach a broader audience for their generative AI-powered features, which enables them to create richer and more intelligent mobile experiences that focus on user needs spread across multiple brands and devices.

App developers interested in adding on-device generative AI to their Android apps face significant challenges in today’s environment. The experimental AI Edge SDK from Google enables developers to utilize the Neural Processing Unit for AI model execution, yet remains restricted to Pixel 9 devices while primarily handling text processing, which limits its broad application. Qualcomm and MediaTek provide proprietary APIs for AI workload management, although their varying feature sets across chipsets and devices complicate their use for sustained development projects. Custom AI model development and implementation require deep knowledge of generative AI systems due to their complex nature. These new APIs, which utilize Gemini Nano as their base, enable developers across various skill levels to access local AI functionalities more efficiently while simplifying implementation processes, which will drive innovation in mobile application development.