google-ai-edge/litert (LiteRT) is Google's on-device ML/GenAI deployment framework that succeeds TensorFlow Lite. It bundles model conversion, runtime, and optimization into a single flow to run machine learning and generative AI on edge devices. The latest release of this repository is v2.3.0, sourced from the GitHub repository google-ai-edge/litert. However, the given records contain no release notes for v2.3.0, so what changed in this version is unknown.
The framework is a lightweight inference runtime for on-device AI, providing consistent APIs for GPU/NPU (neural processing unit) acceleration and model execution on mobile and edge. It is written in C++ and released under the Apache-2.0 license. It supports CPU execution alongside GPU/NPU acceleration, offers the Compiled Model API, and guides on-device and web inference paths through LiteRT-LM and LiteRT.js. The repository has 3507 stars.
There is not enough basis to judge. The records contain no performance benchmarks or per-model runtime and memory figures. The score fills in only 2.8 momentum out of 10, with the freshness, technology, and quality items left empty. The project was created on September 4, 2024, so there is little basis for gauging its maturity and activity scale. The changes in v2.3.0 can only be known by reading the release notes directly, and it makes sense to keep it as a candidate until actual deployment cases are confirmed.