llama.cpp can quantize a model for you:

https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#qu...