Stockfish recommend me to play an inaccuracy ?

<iframe src="lichess.org/embed/iuHxVwOJ?theme=auto&bg=auto"
width=600 height=397 frameborder=0></iframe>

So I was taking a look at my game, as I'm trying to perfect my Modern Defense (I'm playing black)
So this is what happen :
1.e4 g6
2.d4 Bg7
3.Nc3 d6
4.Be3Nf6
5.Be2?!
and here :
...O-O?!

Castling is an innacuracy. So i check what stockfish recommend, and he tells me to castle and do the innacuracy. What is this ?
Anybody would be able to explain ?

Aside from that, it was a boring game with a few blunder imo.
I would appreciate if you see something interesting though !

Just ignore engine evaluations in the opening. The 32 men position is too complex with too many possibilities for any meaningful evaluation within the calculation horizon.
The comment that 5...O-O would be an inaccuracy is probably with lower depth than the recommendation to play 5...O-O.
5...O-O and 5...c6 are good moves.

Alright, c6 seems like a good move, I might play this instead then.
Thanks !

just wait until someone invents a 32 man tablebase ;)

For me it says that O-O is inaccuracy, best move was O-O :D

Stockfish you are drunk, go home

Yes, O-O is +1.02 for white, whereas O-O is +1.21 for white., which is why O-O is superior to O-O.
Hope this helps @nchebykin :)

I guess the difference comes from the evaluation methods. The search for inaccuracies uses a different (greater?) depth, which usually goes unnoticed, but in rare cases can lead to silly situations like this.

It would be interesting to know if SF at different depth suggests a better move.

Even though I can't be sure about the reason, I don't think it is supposed to give you this inaccuracy.

I posted the bug in the feedback forum

lichess.org/forum/lichess-feedback/stockfish-recommends-move-and-calls-it-inaccuracy-at-the-same-time