[timerol]

"All models are wrong. Some models are useful." is the way that I have heard this thesis quipped. I suppose "All models are wrong. Some models are more wrong than others." would fit Asimov's points better, but I've never actually heard that one

[procaryote]

Also, some models aren't even wrong, because they fail to make testable predictions

[toolslive]

"not even wrong"

https://en.wikipedia.org/wiki/Not_even_wrong

[nyrikki]

The map-territory relation is possibly a better fit?

More specifically how Alfred Korzybski put it: "The map is not the territory".

The map will never be the territory, and we are just looking for better maps.

[nemosaltat]

My father-in-law, something of a renaissance man, and also a long-time successful commodities trader, regularly uses your latter expression. The first time I heard him say it, I attempted to correct him (“some are useful”) and he confidently brushed me aside. As I’ve matured, I’ve grown confident that his (your) version is less wrong.