Google Open-Sources ML Drift for Cross-Platform GPU-Accelerated On-Device AI Inference

Original sourceInfoQ

Summary

Google has released ML Drift, an open-source, cross-platform GPU compute engine for on-device AI inference that abstracts different GPU hardware and low-level APIs such as OpenGL ES, OpenCL, Metal and WebGPU, and serves as the foundation for LiteRT. Compared with the older TensorFlow Lite GPU deleg…

Key points

Developers deploying inference on phones or desktops can learn the TFLite GPU delegate retirement timeline and the migration path to its replacement.
It replaces the long-standing TFLite GPU delegate, affecting the inference stack and hardware compatibility strategy of many on-device AI applications.
Teams still relying on the old TFLite GPU delegate must plan migration, or future compatibility and performance optimization will be limited.

AI Tools Weekly

Editorial note

This page is Code & Chain's editorial summary of public sources. It may be prepared with AI assistance and published through an automated workflow. Refer to the original sources; this content is not investment, legal, or tax advice.