LiteRT Adds Faster GPU Inference in Google Play Services
Google announced an upgrade to LiteRT in Google Play services, bringing ML Drift-powered GPU inference to the new CompiledModel API. Google says GPU inference runs about 1.4 times faster than the legacy GPU delegate, while apps that remove their bundled runtime are about 8 MB smaller in its testing. The legacy GPU delegate and custom delegate APIs are deprecated, and NNAPI is scheduled for removal starting in 2027.
Developers running custom models on Android can use a shared, automatically updated inference runtime instead of bundling LiteRT, reducing app size while adding modern GPU acceleration.
Smaller apps and faster GPU inference with LiteRT in Google Play services - Google Developers Blog
Smaller apps and faster GPU inference with LiteRT in Google Play services
OCT. 28, 2026
#Google#LiteRT in Google Play services#ML Drift#CompiledModel API#Google Play services#Android#TensorFlow Lite#Android Studio