Google has officially graduated LiteRT into production, delivering up to 1.4x faster GPU performance and enhanced NPU acceleration for high-performance on-device AI. This update streamlines cross-platform deployment for models from PyTorch, JAX, and Keras while reinforcing the TensorFlow ecosystem with improved security and dependency support.

GoogleDevs' platform is a central hub for developers interested in Google technologies, APIs, and developer tools. Through articles, tutorials, and documentation, GoogleDevs offers insights into building applications using Google Cloud Platform, Android, Chrome, and other Google services. Developers can learn about cloud computing, machine learning, and mobile app development with Google's developer tools and platforms.

Google Developers

TensorFlow 2.21 is released, with the headline change being LiteRT's graduation into full production status. LiteRT is Google's high-performance on-device inference runtime, offering up to 1.4x faster GPU performance and enhanced NPU acceleration compared to TFLite, which it replaces. It supports cross-platform deployment for models from PyTorch, JAX, and Keras. The TensorFlow team also commits to faster bug fixes and more timely dependency updates going forward, while recommending Keras 3, JAX, and PyTorch for new generative AI work.

What's new in TensorFlow 2.21