Llama.cpp b9861 drops with cpp-httplib 0.49.0 and expanded platform support
New release adds KleidiAI for Apple Silicon, CUDA 13, and Windows arm64 OpenCL Adreno
Deep Dive
ggml-org released llama.cpp b9861, updating cpp-httplib to 0.49.0.
Key Points
- Updated cpp-httplib dependency to v0.49.0 for improved networking
- Added KleidiAI acceleration for macOS Apple Silicon (arm64)
- New Windows arm64 build with OpenCL Adreno and CUDA 13 support
Why It Matters
Broader hardware support and library updates make local LLM inference more accessible on consumer devices and servers alike.