llama.cpp b9608 releases with cpp-httplib 0.47.0 upgrade

The open-source llama.cpp project, known for running large language models locally on consumer hardware, has tagged release b9608. The most notable change is the update of the bundled cpp-httplib library from an earlier version to 0.47.0, which brings improved HTTP request handling and security fixes. This library powers the built-in API server that allows users to interact with models via HTTP endpoints, so the update translates to more stable remote inference.

The release also expands pre-built binary support across platforms. macOS users get both Apple Silicon and Intel builds (with KleidiAI optionally enabled for ARM64). Linux builds cover x64, ARM64, and s390x architectures with Vulkan, ROCm 7.2, and OpenVINO backends. Windows users can download x64 or ARM64 builds with CUDA 12/13, Vulkan, or HIP support. Android ARM64 and openEuler builds are also available. This ensures broader device compatibility for local LLM deployments.