Chess.com's Stockfish 16 Lite plays differently than the one from the Stockfish website. Huh?

So, I was analysing a game using Chess.com's Stockfish 16 Lite, and, for some reason, I just randomly decided to compare it to the Stockfish 16 Lite from the actual website. Same depth (99 half moves), same strength (maximum), same engine (I made sure in the settings), but both disagreed on the first move to play. Chess.com's SF16L said to play [1. e4], while SF16L from the Stockfish website said to play [1. b3], [1. Nf3], or [1. d4] instead. Strange.

For another example, Stockfishchess.org's SF16L said that the Polish Opening [1. b4], resulted in no advantage for either side, with a 0.0 rating. Chess.com's SF16L said that [1. b4] resulted in Black having a -0.3 advantage.

Stockfishchess.org's SF16L says that the best four starting moves for a game are [1. b3 d5 2. Nf3 Bf5], while Chess.com's SF16L says that the old [1. e4 e5 2. Nf3 Nc6] are the best to play.

From what I found, it looks like Chess.com's SF16L says to play a more classical, [1. e4 e5] kind of style, while Stockfishchess.org's SF16L says to play a more hypermodern style.

Someone please explain why, and which one's better, because this is confusing.

They're both favourited tabs that I have shortcuts to. I don't plan on cheating at all with either.

I only analyse games after I play. If you check my accuracy, it's really low because of how laggy my Chromebook gets. I don't cheat.

It's also really easy to get caught and banned. Not so much of a "great idea", in my opinion. If I'm really into chess, I would actually learn it, not let something else do it for me. And that's what I'm doing, having fun. Not cheating.