When you start trying out local AI on an RTX 3080 Ti, you tend to think, "I want to run the largest model possible" at ...