llama.cpp b9561 syncs ggml with expanded platform support

The latest release of llama.cpp, b9561, focuses on syncing the ggml library, the core tensor computation engine behind the project. This maintenance update ensures that all platform-specific builds—spanning macOS, iOS, Linux, Windows, and Android—leverage the most recent ggml improvements. The release includes a staggering array of pre-built binaries: from macOS Apple Silicon (arm64) with optional KleidiAI acceleration to Windows with CUDA 12/13, Vulkan, and HIP support, as well as Linux variants for CPU, Vulkan, ROCm 7.2, and OpenVINO.

For developers and enthusiasts running LLMs locally, this update guarantees compatibility with the latest ggml optimizations without requiring manual compilation. The sync also covers edge cases like Android ARM64 CPU and openEuler with ACL Graph, showcasing the project's commitment to broad hardware support. While no major new features are announced, this release is crucial for maintaining performance and stability across the diverse ecosystem of local LLM inference. Users can download the appropriate binaries from the GitHub releases page and continue using llama.cpp as a reliable backend for models like Llama, Mistral, and Gemma.