Google Tensor G3 brings advance on-device machine learning to Pixel 8 series


With the all-new Pixel 8 series, Google has unveiled its 3rd generation custom chipset, Tensor G3. This custom silicon chip brings numerous AI advancements and cutting-edge capabilities, aiming to bring the transformative power of AI to your everyday life, as claimed by Google.

On-Device Machine Learning 

The Tensor G3 chip marks a significant milestone in Google’s journey to advance on-device machine learning. It features a host of upgrades across key subsystems, including the latest ARM CPUs, an enhanced GPU, a new ISP, Imaging DSP, and a next-gen TPU purpose-built to handle Google’s AI models.

Compared to the initial Tensor iteration in Pixel 6, the latest Pixel phones now support more than twice as many on-device machine learning models. These models are not only more numerous but also more sophisticated, promising enhancements across the entire Pixel user experience.

Enabling Generative AI and Enhanced Language Understanding

With the rise of generative AI and collaboration with Google DeepMind, Pixel stands poised to harness the potential of these capabilities. However, on-device generative AI has become incredibly intricate, now 150 times more complex than just a year ago, according to Google. Tensor G3 is up to the challenge, boasting an efficient architecture co-designed with Google Research.

Tensor G3 also takes strides in speech and natural language understanding. It revolutionizes interactions with Google Assistant, now capable of understanding natural pauses and conversational nuances. Pixel 8 introduces the same text-to-speech model as Google’s data centers, enabling webpages to be read aloud and translated seamlessly.

Elevated Photography and Videography

Tensor G3 brings a significant boost to photography and video capture on Pixel 8 and Pixel 8 Pro. The chip’s efficiency optimizes the camera pipeline, enhancing Live-HDR capabilities with improved detail, color accuracy, contrast, and dynamic range while conserving power.

Additionally, computational photography sees notable advancements with features like Magic Eraser, capable of removing distractions in photos, and Best Take, which harnesses advanced machine learning models for impeccable image composition. Audio Magic Eraser further refines video quality by reducing background noise, ensuring crystal-clear audio.

Privacy and Security

The Tensor G3 is equipped with the Tensor security core and works in tandem with the Titan M2 security chip, making it more resilient to sophisticated attacks. Further, the enhanced Face Unlock now meets stringent Android biometric standards, enabling secure access to banking and payment apps.

The Tensor G3 chip ushers in a new era of AI-driven mobile technology, promising enhanced user experiences, photography, and robust security for Pixel enthusiasts.

Additional technical details of the chipset were not officially disclosed by Google, but are expected to emerge in the coming days. This will provide more insight into the real-world performance and technical specifications of the Tensor G3.

According to Google, on the Tensor G3 technology,

Our third-generation Google Tensor G3 chip continues to push the boundaries of on-device machine learning, bringing the latest in Google AI research directly to our newest phones: Pixel 8 and Pixel 8 Pro.


Author: Varun Krish

Varun Krish is a Mobile Technology Enthusiast and has been blogging about mobile phones since 2005. His current phones include the Apple iPhone 13 Pro and Google Pixel 6. You can follow him on Twitter @varunkrish and on Google+ You can also mail Varun Krish