Weird, what issue are you having. Works for me both in ollama and native llama.cpp I am using unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:MXFP4_MOE

→ View original post on X — @rasbt