Llama.cpp supports Vulkan, which is supported by all GPU vendors that care about standards and interoperability. The default should be open and portable APIs,

Llama.cpp supports Vulkan, which is supported by all GPU vendors that care about standards and interoperability. The default should be open and portable APIs, not needlessly furthering a hegemony that is detrimental to us all.