Llama.cpp (B5899 <= Version < B10261)

A heap out-of-bounds write vulnerability in llama.cpp versions b5899 through b10260 allows attackers to achieve memory corruption or potential code execution via maliciously crafted GGUF model files.