llama.cpp Releases Version b11346 with Fix for Qwen Experimental Tests
llama.cpp version b11346 has been released, featuring a fix for experimental Qwen model test suites. The release also provides updated pre-compiled binaries across multiple platforms including Linux, macOS, Windows, and Android. As a routine automated release, it ensures test suite stability for developers working with experimental Qwen model architectures. It maintains continuous delivery of prebuilt binaries for users running LLMs locally across diverse hardware backends. The update specifically resolves pull request #29819 ('qwen4exp: fix tests') without introducing major functional changes or architectural overhauls. Binaries cover hardware backends such as CUDA 12/13, Vulkan, ROCm, OpenVINO, SYCL, and Snapdragon NPUs.
## BACKGROUND
llama.cpp is a popular open-source C/C++ inference framework created by Georgi Gerganov for running large language models locally on consumer devices. Qwen is a family of open-weight large language models developed by Alibaba Cloud, frequently integrated and optimized within llama.cpp.