Gemma 4 QAT: Shrinking Powerful AI for Seamless Mobile Performance

This release represents a significant shift toward local AI by enabling high-performance models like Gemma 4 to run on consumer hardware with minimal quality loss. By integrating quantization directly into training, it paves the way for faster, more private, and memory-efficient AI experiences on everyday devices.