llama.cpp's New Agent Flag Unlocks Autonomous AI – Now on Every Major Platform
llama.cpp now supports AI agents with a new --agent argument across 20+ platforms.
llama.cpp, the highly popular open‑source C++ library for running large language models locally, has released version b9726. The headline change is the addition of a new command‑line argument `--agent`, which enables users to run AI agents directly within the server. This eliminates the need for separate workarounds and aligns with the growing trend of deploying autonomous AI agents on consumer hardware. The update also removes redundant webui naming compatibility, streamlining the codebase and fixing environment variable handling for better cross‑platform consistency.
Platform coverage continues to expand aggressively. This release ships pre‑built binaries for macOS (Apple Silicon and Intel, including a version with KleidiAI acceleration), Linux (Ubuntu on x64 and arm64 with Vulkan, ROCm 7.2, OpenVINO, and SYCL FP32/FP16), Windows (x64 and arm64 CPU, CUDA 12 and 13, Vulkan, OpenVINO, SYCL, HIP), Android arm64, and even openEuler (x86 and aarch64 with ACL Graph). Developers can now deploy llama.cpp across virtually any environment, from edge devices to cloud servers, making local AI inference more accessible than ever.
- New `--agent` argument enables running AI agents directly in the llama.cpp server, simplifying autonomous LLM workflows.
- Support expanded to 20+ platform/builds including macOS (Apple Silicon with KleidiAI), Windows (CUDA 12/13, Vulkan, HIP), and openEuler.
- Project maintains massive community traction with 117k stars and 19.7k forks on GitHub, reflecting its dominance in local LLM inference.
Why It Matters
llama.cpp’s agent support and platform breadth make local AI agents viable for professionals seeking privacy and low latency.