llama.cpp b9264 brings broader platform support and version display

The llama.cpp project by ggml-org shipped b9264, its latest release of the lightweight C++ LLM inference engine. This version introduces a simple quality-of-life improvement: the `app : show version` command, allowing users to quickly verify their build number. More significantly, the release expands the already vast platform matrix with new pre-built binaries for Windows arm64, CUDA 12.4 and 13.1, Vulkan, and HIP, plus openEuler Linux on both x86 and aarch64 architectures (including support for Ascend 910b and 310p via ACL Graph). macOS builds are available for Apple Silicon (arm64) with optional KleidiAI acceleration, Intel (x64), and iOS as an XCFramework. Linux users get CPU, Vulkan, ROCm 7.2, OpenVINO, and SYCL FP32/FP16 builds. Android arm64 is also supported.

For developers and power users running LLMs locally, this release ensures broader hardware compatibility without manual compilation. The inclusion of Windows arm64 builds is particularly timely as Snapdragon X Elite and other ARM Windows devices gain traction. The openEuler support targets enterprise Linux deployments, extending llama.cpp's reach into cloud and edge data centers. While not a major feature update, b9264 demonstrates the project's commitment to making local LLM inference accessible on virtually any modern computing platform, from phones to servers.