The llama.cpp project has published version b9823, providing pre-built binaries for macOS, iOS, Linux, Android, Windows, and openEuler platforms. A key change in this release is the addition of a Windows OpenVINO build to the check-release pipeline.

  • macOS Apple Silicon (arm64) builds are available, while KleidiAI support remains disabled.
  • Linux binaries include CPU, Vulkan, ROCm 7.2, OpenVINO, and SYCL variants for x64, arm64, and s390x architectures.
  • Windows offerings now feature CUDA 12.4 and 13.3 support alongside standard CPU, Vulkan, OpenCL Adreno, HIP, and the new OpenVINO builds.
  • Android arm64 (CPU) and iOS XCFramework binaries are included in the release assets.

This update allows users to access the latest llama.cpp features across a wide range of hardware configurations and operating systems.