Help to understand a move in the game between Judit Polgar and Henrique Mecking

Dear chess friends on Chess.com,

Please, I need help understanding a specific move that (was not) played in a game.

The mentioned game, which took place in 2012, was played by GM Judit Polgar (playing as White) and GM Henrique Mecking.

Here's the game:

Note that, on the tenth move of the game, White preferred the move 10. exf5 (capturing the pawn) instead of the move 10. gxf6 (capturing the knight and equalizing in material). I analyzed several lines of play using Stockfish and couldn't find a tangible advantage for White in capturing the f5 pawn instead of the f6 knight.

What did White see? Why wasn't the knight captured?

I thank anyone who can help me. I've been analyzing for days and can't understand it.

But would this advantage be enough to justify being down a piece and missing the opportunity to equalize in material?

@4

White has full compensation for the piece, but no more.
It is even the safer option, as 10 gxf6 puts white as a disadvantage.

But taking the knight on f6 wouldn't change anything for Black's king. It would still be unable to castle. White would still have this advantage...

I saw the games you posted, but it was still not clear why White would be at a disadvantage capturing the knight on f6.

Engines are not good at evaluating positional sacrifices, at least not unless they get enough time to analyze.

Also, the position is good for Judit ( her style is wild and aggressive), while uncomfortable for the elderly grandmaster.

@8

The first game of 2013 shows how black can defend and hold the draw with 12...Nbd7 instead of 12...Qc7? what Mecking played in 2012.

The second game, an ICCF correspondence game of 2006 shows how black gets the advantage after 10 gxf6? instead of 10 exf5.

Thanks for helping me understand some concepts. I still didn’t get much, but I think I’m on the right path.

Thanks to everyone who posted and tried to help me understand this move. I still don't totally understand what White saw, but I believe I will understand after continuing my studies.

@11

This is all accumulated knowledge.
In 2006 white played 10 gxf6 in the correspondence game, but lost horribly.
In 2012 both Mecking and Polgar were aware of that correspondence game and/or other games, so Polgar did not play 10 gxf6, but 10 exf5. Mecking played 12...Qc7? and lost horribly.
In 2013 the players were aware of that game and thus black improved with 12...Nbd7 and held.

Open Sicilians aren't always ABOUT material.

One of my OTB games, from the 1980s:

It might just be theory. She would have studied that particular line through multiple games and decided that in OTB tournaments where her opponent was unprepared that was the most challenging move. Both psychologically, and chess-wise. You might notice that Polgar wins a tempo on the knight, and has a much more active position, with multiple open files and diagonals that she can use to attack. It would be extremely uncomfortable for black to play here. Also, you'll notice that black sacrificed the knight back on the next move, and both sides were continuously sacrificing material over the next 6 or 7 moves.

This is the very sharp Perenyi attack, where white promptly sacrifises a piece (in some lines even two of them!) for very strong pressure.

10.exf6? is a bad mistake because of 10...f4! when the e3 bishop is attacked, and the f6 pawn is very weak (actually Black need not capture it immediately, and keep on developing like ...Be6, ...Nd7 etc).