HF does this for ggufs, and it’ll show you what quantizations will work on the GPU(s) you’ve selected. Hopefully that feature gets expanded to support more model types.