The way I see it, true gambits trade material for position, which in turn could give you opportunities to regain more material. However, in blitz play, you are less likely to notice those opportunities in order to exploit them. Is it fair to say that blitz changes the relative value of position vs material? In particular, should you prefer to accept (and not offer) gambits more?
It just seemed obvious to me that gambits would work better with faster games, for the same reasons as given in the comments - it's harder to defend than attack, etc. I tried looking at several gambits in the lichess database to support this theory.
In classical+rapid games, White wins 53% of Evans Gambit games vs 43% for Black. In bullet+blitz games, it's 51% for White and 46% for Black.
Let's look at the King's Gambit. In classical+rapid games, White wins 53% and Black wins 43%. In bullet+blitz, it's 49% for White and 47% for Black.
Halloween Gambit? Classical+rapid leads to 52% White 43% Black. Bullet+blitz leads to 52% White 44% Black.
Englund Gambit (a gambit for Black) has 49% White 46% Black for classical+rapid, and 50% White 46% Black for bullet+blitz.
As a control I then looked at 1.e4 to make sure this wasn't something that happened in general. 1.e4 had 49% White and 46% Black for both classical+rapid and blitz+bullet, so this isn't simply a general effect of the time control getting faster.
Well then. This kind of busts my theory. It seems the gambiting side has a better record in the slower time controls, and the defending side has a better record in the fast time controls, after all! The effect is, however, small, and probably not worth basing your play on.