Topic: native model execution

Google’s Gemma 4 optimized for Pixel 10’s TPU

Google launched Gemma 4 E2B, an AI model fine-tuned for Tensor Processing Units (TPUs) that runs directly on the Pixel 10’s custom TPU for faster, more power-efficient on-device inference. This on-device execution reduces reliance on cloud servers, improving latency and privacy while enabling tas...