I've used their gemma 4 quants since when they were not still working in llama.cpp and ik-llama.cpp and I don't remember any problems

They are the most reliable in my experience, but if you have alternatives you trust I'd love to know