ggerganov/llama.cpp Released Version 0.7.0 (Build b11669)
Llama.cpp has released version 0.7.0 (build b11669), which bumps the library version and provides precompiled binaries for a wide variety of operating systems and hardware architectures. This release streamlines the deployment process for developers and enthusiasts running large language models locally by offering ready-to-use binaries across diverse platforms including macOS, Linux, Windows, and Android. The release includes targeted builds for various hardware accelerators such as CUDA, Vulkan, OpenVINO, SYCL, and Snapdragon NPUs, alongside standard CPU implementations.
## BACKGROUND
Llama.cpp is an open-source software library co-developed alongside the GGML project that performs inference on large language models like Meta's Llama locally. It serves as the core inference engine for many popular local AI tools, including Ollama and LM Studio.