llama.cpp b9405 release optimizes builds, moves licenses to llama-app
The open-source LLM inference tool adds new build options and cleans up licensing
Deep Dive
ggml-org's llama.cpp released version b9405 on May 29, with a commit that moved licenses to the llama-app directory, signed by Adrien Gall
Key Points
- Licenses moved to llama-app directory to improve project structure and clarity for contributors
- Pre-built binaries available across 20+ platform/backend combinations including CPU, Vulkan, CUDA 12/13, ROCm 7.2, OpenVINO, HIP, and KleidiAI (disabled)
- Release targets macOS (Apple Silicon, Intel, iOS), Linux (x64, arm64, s390x), Android arm64, and Windows (x64, arm64)
Why It Matters
Streamlined licensing and broadened build support make llama.cpp easier to deploy locally for AI inference on diverse hardware.