The ggml-org team has released llama.cpp b12032, with downloadable builds for macOS, iOS and several Linux systems. The release notes list one code change: SYCL builds will try PkgConfig to locate Level Zero include files and libraries. The notes do not describe any changes to model quality or runtime speed.
Why it matters
Developers building llama.cpp with SYCL may have another way to locate Level Zero dependencies; the release notes do not say which setups benefit or whether the change improves performance.
