Google has introduced Gemma 4 E2B, a lightweight yet powerful AI model designed to run natively on the Pixel 10’s Tensor Processing Unit (TPU). This development enhances the Pixel 10 series’ on-device AI capabilities, offering users advanced features without relying on cloud processing.
Gemma 4 E2B is part of Google’s Gemma series of open models, which are engineered for efficient on-device performance. This model is compatible with the Pixel 10, 10 Pro, 10 Pro XL, and 10 Pro Fold, leveraging the Tensor G5’s TPU to deliver high-performance AI functionalities directly on the device.
The model’s multimodal capabilities enable a range of applications:
- AI Chat: Facilitates instant, in-depth offline conversations, even in environments without internet connectivity.
- Ask Image: Allows users to capture and identify objects, plants, or issues without requiring an internet connection.
- Ask Audio: Provides private, on-device audio transcription for lectures and notes, ensuring user data remains secure.
Additionally, Gemma 4 E2B introduces Mobile Actions, enabling users to control core phone functions—such as Wi-Fi settings or navigation—using voice or text commands processed entirely on-device. This feature enhances user convenience while maintaining privacy.
Real-world applications of Gemma 4 E2B include:
- Retail: Transforming recipe ideas into precise, localized in-store shopping maps, all processed offline.
- Automotive: Providing mechanics with immediate visual diagnostics from photos of faulty parts, streamlining repair processes.
By integrating Gemma 4 E2B into the Pixel 10 series, Google underscores its commitment to advancing on-device AI capabilities. This approach not only enhances user experience through faster and more reliable AI features but also prioritizes user privacy by minimizing data transmission to external servers. As on-device AI continues to evolve, we can anticipate more personalized and efficient interactions with our devices, setting a new standard for mobile technology.
