The latest release of llama.cpp, build b10227, introduces a specialized parser for the Qwen3 model family, enhancing how the chat interface handles tagged thinking tools. This update refactors internal chat processing and adds a permute helper, streamlining the codebase for future expansions. Self-hosted users will benefit from more accurate tool calling, as the update adds support for omitting certain tool call tags and adjusts trigger patterns to ensure functions like <function are properly recognized. These changes, alongside updated tool delimiters and a note for the Qwen3-Coder variant, collectively improve the reliability of interactive chat sessions and automated function calls when running Qwen3 models locally.
In addition to the parser improvements, the release provides downloadable binaries for a wide range of platforms, ensuring that users can deploy llama.cpp on their preferred operating system and hardware. Below is the complete list of available builds:
Website:
macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) — currently disabled
- macOS Intel (x64)
- iOS XCFramework
Linux:
- Ubuntu x64 (CPU)
- Ubuntu arm64 (CPU)
- Ubuntu s390x (CPU)
- Ubuntu x64 (Vulkan)
- Ubuntu arm64 (Vulkan)
- Ubuntu x64 (ROCm 7.2)
- Ubuntu x64 (OpenVINO)
- Ubuntu x64 (SYCL FP32)
- Ubuntu x64 (SYCL FP16)
Android:
Windows:
- Windows x64 (CPU)
- Windows arm64 (CPU)
- Windows arm64 (OpenCL Adreno)
- Windows x64 (CUDA 12) — CUDA 12.4 DLLs
- Windows x64 (CUDA 13) — CUDA 13.3 DLLs
- Windows x64 (Vulkan)
- Windows x64 (OpenVINO)
- Windows x64 (SYCL)
- Windows x64 (HIP)
openEuler:
- Currently disabled — openEuler builds are not available at this time
- openEuler x86 (310p)
- openEuler x86 (910b, ACL Graph)
- openEuler aarch64 (310p)
- openEuler aarch64 (910b, ACL Graph)
UI:
