llama.cpp b9909 adds timings and progress to /responses API stream

The llama.cpp project, one of the most popular open-source tools for running large language models locally, has released version b9909. This update introduces timings and progress information to the server's /responses API stream, allowing developers to track inference speed and generation status in real time. The feature is particularly valuable for building responsive user interfaces, monitoring performance, and debugging latency issues when serving LLMs.

Version b9909 is built from commit 5eca4e3, signed by GitHub with verified signature. The release provides prebuilt binaries for a wide range of platforms and configurations, including macOS (Apple Silicon with and without KleidiAI, Intel), Linux (x86, ARM, s390x with CPU, Vulkan, ROCm 7.2, OpenVINO, SYCL), Windows (CPU, ARM64, OpenCL Adreno, CUDA 12/13, Vulkan, OpenVINO, SYCL, HIP), and Android ARM64. Some builds like openEuler and KleidiAI for macOS are marked as DISABLED in this release. With 120k stars and over 20k forks, llama.cpp remains a cornerstone for local LLM inference, and this incremental update improves developer experience for API-based applications.