llama.cpp b9103 released with updated HTTP library and broad platform builds
The 110k-star local LLM runtime gets a maintenance release with cpp-httplib 0.44.0.
Deep Dive
ggml-org released llama.cpp b9103, updating the cpp-httplib dependency to 0.44.0. The release includes builds for macOS (Apple Silicon, Intel, iOS), Linux (Ubuntu x64/arm64/s390x with CPU, Vulkan, ROCm 7.2, OpenVINO, SYCL FP32/FP16), Android (arm64, CPU), Windows (x64 CPU, arm64 CPU, x64 with CUDA 12/13, Vulkan, SYCL, HIP), and openEuler (x86/aarch64 with 310p/910b ACL Graph).
Key Points
- Updated cpp-httplib from previous version to 0.44.0 for better HTTP handling
- Prebuilt binaries for macOS, Linux, Windows, Android, iOS, and openEuler across CPU, CUDA, Vulkan, ROCm, OpenVINO, SYCL, and HIP
- Includes specialized builds like macOS with KleidiAI acceleration and Windows with CUDA 12/13 DLLs
Why It Matters
Up-to-date dependencies and broad platform support keep llama.cpp viable for local AI inference across diverse hardware.