Model authors are welcome to add support to llama.cpp before release like IBM did for granite 4 https://github.com/ggml-org/llama.cpp/pull/13550