Yep, the line involved white giving his e-pawn though so I assumed it was a win. Verified now with variations since the position was bugging me
here are some lines showing that the position after 1.Qxg4 hxg4 is -+ and Black even has a little wiggle room (analysis courtesy of me + houdini):
2.Kg3 b5?!
(2.-Ke7/Kf7/a5 all seem to win more easily, for example
2.-Kf7 3.Kxg4 Bxg2 4.h5 Be4 5.a4 a5 or
2.-a5 3.Kxg4 Bxg2 4.h5 Be4 5.a4 Kf7 6.Kg5 Bc2 7.h6 Bxa4 8.e6+ Kg8)
3.Kxg4 Bxg2
4.Kf4 Kf7! (not 4.-a5 5.a4! and draws)
5.a4 b4! 6.cxb4 Bh3! heading for d7 with a win
2.a4 a5!
3.Kg3 Bc2 (3.-Bf5 is maybe even simpler and was what I looked at earlier, the engine wants to play 4.Kf4 Bd7 5.e6 Bxe6 but this looks pretty easy for Black to win)
4.Kxg4 Bxa4 5.Kf4 (not 5.Kg5 b5) Kf7
6.Ke3 Kg6
7.g4 is close to a fortress but not quite: 7.-Bd7
8.Kf4 b5
9.a3 Kg7 (probably there are quite a few winning squares for the king)
10.Kf3 a4!
11.Kf4 Bxg4
12.Kxg4 b4
13.h5 and now both bxa3 and bxc3 win for Black