ggml‑org released llama.cpp version b12088. The update enables the NVFP4 LFM3 model format and ships pre‑built binaries for Apple Silicon, Intel macOS, iOS, and multiple Ubuntu configurations, including CPU‑only and Vulkan builds. The macOS Apple Silicon build with KleidiAI is listed but currently disabled. All files are available for download from the GitHub release page.
Why it matters
Developers can now run llama.cpp with the new model format and have ready‑made binaries for more platforms, reducing setup time.
