The ggml‑org team released llama.cpp version b12531. The update adds support for the voxcpm2 model format and a slight speed improvement. It also includes fresh pre‑built binaries for macOS Apple Silicon, macOS Intel, iOS, and several Ubuntu builds. The changes are listed as “mtmd: support voxcpm2” and a handful of fixes.
Why it matters
Developers can now run the newer voxcpm2 models on more platforms with a small speed gain.