Google AI Edge is a full-stack platform for deploying machine learning and generative AI models on mobile, web, and embedded devices, enabling low-latency, offline, and privacy-preserving inference at the edge.[2][3] It provides cross-platform support so the same model can run on Android, iOS, web, and microcontrollers, with compatibility for JAX, Keras, PyTorch, and TensorFlow models through components like LiteRT.[2][4] Developers can use low-code MediaPipe Tasks APIs for common vision, text, audio, and generative AI workloads, or build custom accelerated ML pipelines via the MediaPipe Framework.[2] Tooling such as Model Explorer helps visualize model transformations, optimize quantization, and debug performance hotspots across CPU, GPU, and NPU hardware accelerators.[2][4] The stack powers many of Google’s own on-device AI experiences and is designed for scalable, production-grade edge AI deployments.[2][3]