Do you have a GPU with enough vRAM to fit it? Or are you just running on CPU?

Also, if you’re in that size range already, consider the Dolphin line of models from cognitive computations. They’re trained to be unaligned and maximally compliant with user requests.