llama.cpp releases version b9751 with updated binaries for macOS, Linux, Android, Windows, and openEuler. The release includes fixes for memory usage and supports multiple architectures and hardware acceleration options like Vulkan, CUDA, OpenVINO, and SYCL.