llama.cpp Releases Version 0.3.0 (Build b10621)
The llama.cpp project has officially bumped its version to 0.3.0 (build b10621). This release updates the project's versioning scheme, improves continuous integration (CI) release descriptions, and adds scripts for generating release summaries. As a foundational open-source engine for local LLM inference, version bumps in llama.cpp signal milestones in its lifecycle and release automation. This update helps streamline deployment pipelines for developers and platforms that rely on llama.cpp for local AI applications. The release provides pre-compiled binaries for multiple platforms including Windows, Linux, macOS, Android, and iOS, supporting backends like CUDA, Vulkan, ROCm, and OpenVINO. However, certain builds, such as the macOS Apple Silicon build with KleidiAI enabled, remain disabled in this release.
## BACKGROUND
llama.cpp is a high-performance, open-source inference engine written in C/C++ that allows users to run large language models locally, serving as the core engine for popular tools like Ollama and LM Studio. It is co-developed alongside the GGML tensor library. KleidiAI is a library of micro-kernels optimized for the Arm architecture to accelerate AI workloads.