Docs: Clarification request: Which engines can run these?

[lee-b]

It's not entirely clear:

A novel MoE-aware mixed-precision quantization technique for llama.cpp

Can llama.cpp run these already? It this just a quantization tool that applies existing GGUF capabilities more optimally, or is it a new/enhanced format on disk or in memory?

[cemalgnlts]

I managed to run it with Llama.cpp—well, that is, if it's a model that has apex-quant. Though I'm not entirely sure if this is enough to prove that it's supported."