The llama.cpp project has released version b10045, introducing a new feature to the server component that allows saving and restoring slots using text-only formats via mtmd.
This update is available across multiple platforms and hardware configurations, including macOS (Apple Silicon and Intel), iOS, Linux (Ubuntu x64, arm64, s390x with CPU, Vulkan, ROCm 7.2, OpenVINO, and SYCL backends), Android (arm64 CPU), Windows (CPU, OpenCL Adreno, CUDA 12/13, Vulkan, OpenVINO, SYCL, HIP), and openEuler.
The release provides binaries for the core server and UI, enabling users to deploy llama.cpp on a wide variety of architectures and operating systems.