llama.cpp b9763 adds ID to tool call API responses
New release improves tool calling with unique IDs for each response
llama.cpp, the popular open-source project for running large language models locally, has released version b9763. The standout feature is a new API improvement: the server now includes unique IDs in tool call responses. This addresses a key developer need for matching function call requests to their corresponding results, critical for building reliable agent workflows and multi-turn tool-using applications.
The release also continues llama.cpp's tradition of broad platform support. Updated builds are available for macOS (Apple Silicon and Intel), Linux (x64, arm64, s390x with Vulkan, ROCm 7.2, OpenVINO, SYCL), Windows (CPU, CUDA 12/13, Vulkan, OpenCL Adreno for arm64), and Android. macOS users also get builds with KleidiAI optimizations enabled. This release, tagged b9763, is signed with GitHub's verified GPG key for security.
- Server API now includes an 'id' field in tool call responses, improving traceability for function-calling workflows
- New builds for multiple backends: CPU, CUDA 12/13, Vulkan, ROCm 7.2, OpenVINO, SYCL, and OpenCL Adreno
- Includes experimental builds for openEuler with ACL graph optimizations, plus KleidiAI on macOS Apple Silicon
Why It Matters
Enables developers to build more reliable AI agents using local LLMs with proper tool call tracking.