Self-contained C++ library Efficient inference on consumer CPU and GPUs (NVIDIA, AMD, Intel) Lightweight deployment on many platforms (Windows, Linux, MacOS) Growing number of supported models behind ...