Suboptimal Move Suggestions in Game Review

Hi, I have noticed that the game review is suggesting suboptimal moves on several occasions despite the best moves being fairly obvious. 

In the position below (open image in new tab if it's blurry to you), our dark square bishop is attacked by black's pawn, since we are superior in terms of material and there is no immediate tactic that we can utilize other than moving our bishop to recapture the g pawn, Bxg5 (with a threat of Rd8+ then Bf4 pin if the black rook ever tries Rxe5 to regain their pawn) should be the best move.

However, as you can see below, the game review engine (at 30 depth) is suggesting that Bxg5 is just a good move (not even an excellent alternative move) and wants us to simply sacrifice our bishop cleanly when there is absolutely no need to do so.

In fact, self-analysis at depth 20 through 30 all suggests that Bxg5 is the best move in all the engines on chess.com.

I believe that the game review is either using a suboptimal chess engine than what's listed or the actual depth is not as high as what we pick.

I think it shows the bishop sac as "Best" because the pawns on e4 and f4 are unstoppable and black has to sac a lot of material to stop them.

When posting analysis issues it would be best to include a link to the game so the situation came be investigated.

Yeah, a link to the analysis or game would be nice, or the PGN.

I'm testing on my phone on the start position, and got an interesting result, but want to try on my computer.

OK, when I tested on my phone earlier, I got that Kd3 was the best move.

Putting it on a local install of Stockfish 12 (stock UCI options):

d26/7: gives Nd4

d30: gives Kd3

So, I don't really think there is any problem. It all comes down to depth. Evaluations are close and any run may have a slightly different value at a given depth.

Stopped at d37 locally:

And here is the browser based version at d35 (being in the browser and Javascript likely has some bearing in the results):

In this game, mate in 1 seems like a fairly optimal suggestion on moves 20 and 21.

https://www.chess.com/game/live/30285588771