# LiteRT

[LiteRT](https://github.com/google-ai-edge/LiteRT) (Lite Runtime) is the
successor to TensorFlow Lite and Google’s high-performance runtime for
on-device AI. On Qualcomm platforms, the LiteRT Qualcomm integration connects
the LiteRT model deployment and execution workflow to the QAIRT acceleration
stack, enabling supported models to leverage Qualcomm hardware acceleration
for inference using the applicable backends and runtime libraries.

As Google has stated that future feature updates and performance enhancements
will be exclusive to LiteRT, developers building new on-device AI applications
within the TensorFlow ecosystem are recommended to use LiteRT with the Qualcomm
integration as the replacement for the legacy QNN TFLite Delegate path.

For setup, build instructions, examples, and latest usage guidance,
see the
[LiteRT Qualcomm home page](https://github.com/google-ai-edge/LiteRT/blob/main/litert/vendors/qualcomm/README.md).

Last Published: Sep 03, 2026

[Previous Topic
qnn-genai-transformer-composer](https://docs.qualcomm.com/bundle/publicresource/80-63442-10/topics/qnn-genai-transformer-composer.md) [Next Topic
TFLite Delegate](https://docs.qualcomm.com/bundle/publicresource/80-63442-10/topics/index_TfLite.md)