I dont often get the evans gambit for some reason i think its the only sound one of those three
Evans gambit vs Kings gambit vs Danish gambit
Yeah the danish and kings gsmbit is fine and i play them myself:) just find the low popularity of the evans gambit along my opponents fascinating as i think theoretically its the only one were i think white has full compensation for the pawn

I love the evans gambit as white. If black plays correctly he can easily equalize, but if he is careless he soon has an army of white pieces aiming directly at his king.

The Evans gambit is supposed to be refuted (but at club level it makes no difference because nobody knows the refutation). I don't really know the Danish gambit, although it sounds dubious to me (Larsen played it, and Larsen made a living of playing dubious openings). It seems that white has not much more than an equal position with the King's gambit, but yet again it is fine at club level (but the thrill of playing 2 f4 is worth it).
The Budapest gambit and the Gambit Benko seem to be two strong and sound gambits for black, and they are probably sounder than the three already mentionned.

When was Evan's supposedly refuted? It was recently played against Caruana(can't remember who was White), I don't think its refuted...

The Danish is not sound. Believe me, I love the Danish, but any player over 2200-2400 that plays the Danish against a similarly rated player is going to get clobbered. It's a pure pawn down opening, and good players head to head will usually end up that way, with black winning by nursing the extra pawn they retain to a win (after giving one back somewhere along the way and consolidating their position). The opening position gained is well worth sacrificing 1 pawn, but not 2.

I was impressed that Kasparov played the Evan's against Anand in 1995. And blew Anand away in 25 moves. A fitting tribute in a memorial tournament for Tal.

When was Evan's supposedly refuted? It was recently played against Caruana(can't remember who was White), I don't think its refuted...
Well, it ws Nisipeanu-Caruana in Dortmund, and Caruana won that game, so it is certainly not a proof that the Evans is sound.
Lasker proposed a refutation by giving back the pawn (look at the game by Caruana : black gives back a first pawn, then a second for better structure and pieces activity). I don't know what modern theory think of it, but since it is so rarely played at high level it seems it is considered dubious there.
I've also forgotten the Marshal Gambit in the Ruy Lopez. When an actual line of an opening is called "anti-Marshall", it says something about the soundness of the gambit.

high level gms do still play the evans occasionally and their performance isnt so bad with it. We shouldnt forget that Caruana is rated much higher than nissipeanu so the result is rather predictable, and may have had ntohign to do with the opening. Nissipeanu however is not weak enough to play refuted openings. (not even in larsen style...he plays legit stuff).
Its pretty common for people to assume that because 2700+ rarely play an opening it must be dubious. This simply isnt necessarily true.
Jobava. Rapport.
Besides Nisipeanu is only rated 100 points behind Caruana, that's not so big a difference, particularly with white.
Did you look at the game ? Obviously I'm not good enough to say that with any kind of confidence, but it seems to me that Nisipeanu got nothing out of the opening, which is a bad sign if you are white...
100 points isnt much difference in skill at clublevel but its a huge skilldifference between for example Carlsen and Gelfand which is 100 points. The higher the rating the more they matter

100 points isnt much difference in skill at clublevel but its a huge skilldifference between for example Carlsen and Gelfand which is 100 points. The higher the rating the more they matter
How do you figure this?
FIDE ratings are based on the Elo system and in the Elo system a rating advantage of 100 points means that higher-rated player will statistically score 64% against the lower rated player, no matter how high or low their ratings are.

That wasn't the only game in the past couple months that the Evan's Gambit was played against Caruana:

100 points isnt much difference in skill at clublevel but its a huge skilldifference between for example Carlsen and Gelfand which is 100 points. The higher the rating the more they matter
How do you figure this?
FIDE ratings are based on the Elo system and in the Elo system a rating advantage of 100 points means that higher-rated player will statistically score 64% against the lower rated player, no matter how high or low their ratings are.
It's simple...the closer you are to the top (or bottom) of the Elo ratings pool in question, the harder it is for someone to gain (or lose at the bottom) further points. Carlsen, the ultimate outlier, loses ratings points from every draw. Even if a prodigy came along that could wipe the floor with the best engines and should in theory have a 4000 rating, it just cannot happen, because the pool you are part of drags you down to within 400 points of itself. Thus, currently a 3600ish performance rating is the best you can have, but someday, this will not be true anymore once the entire pool continues to expand and attenuate. The Elo ratings pool is not linear...it has asymptote-ish edges. This is why, for example, Fischer's 120 point ratings edge over the next best player is considered such a big achievement. It's incredibly hard to push the edge of balloon outward all by yourself ...it's like trying to climb into the air when you have reached the tops of everyone's shoulders and have nothing to step on anymore.
Pushing into the outlier area of 2700+ is not the same as going from 1500 to 1600. Every GM in the 2500-2600 range is a ratings landmine you can step on, and such opponents are deadly if you make one positional slip that most chessplayers would not even consider a mistake. So, it is much harder and requires a greater skill differential for a Carlsen to win 64% of the time over a Michael Adams than it is for a weekend tourney club player at 1600 to hold 64% against his blunders-every-game 1500 rated buddy.
100 points isnt much difference in skill at clublevel but its a huge skilldifference between for example Carlsen and Gelfand which is 100 points. The higher the rating the more they matter
How do you figure this?
FIDE ratings are based on the Elo system and in the Elo system a rating advantage of 100 points means that higher-rated player will statistically score 64% against the lower rated player, no matter how high or low their ratings are.
pretty much what btickler said. put a 800 against a 900 and its like playing the lottery
I say Danish is sound since the opponent usually chooses to return the pawns

Litteraly in about 15 moves
Because you are playing someone who doesn't know what they are doing. I urge you to download Stockfish or something, then play the Danish opening position, and have Stockfish play against itself with each side at the highest settings. I guarantee you that white will lose or draw every single game.
Note: I am not saying you get to force Stockfish to accept the full gambit...you need to start the game with only e4 e5 d4 exd4 c3 and run it from there and allow Stockfish to go a full 25-30 ply per move as well. From there you can later try to force certain variations to see if they work, once you prove that the "theoretical best" line is draw or lose. You can try forcing cxb2, and the "book" d5 push for black giving a pawn back, etc.
Trust me, when I did not want to give up the Danish, I did this myself to try and redeem it somehow and prove to myself it was theoretically viable up to rating levels I might aspire to reach. It just isn't.

Litteraly in about 15 moves
Because you are playing someone who doesn't know what they are doing. I urge you to download Stockfish or something, then set up the Danish opening position, and have Stockfish play against itself with each side at the highest settings. I guarantee you that white will lose or draw every single game.
Trust me, when I did not want to give up the Danish, I did this myself to try and redeem it somehow and prove to myself it was theoretically viable up to rating levels I might aspire to reach. It just isn't.
i would agree the danish isnt usually a great choice at higher level but this is something you would figure out on your own with some experience. but having computers play themselves is not a good way to make such a conclusion. computers just do not play like humans...not even 2800 humans. there would be times the computers are playing moves with very obscure long tactical variations in mind well beyond the grasp of some super gm and there would be times the gm could improve on the play of either side for positional or endgame reasons.
Using an engine was just to confirm my long experience ;). But, if you go through my game history you'll see me winning games with the Danish and Bird's (yes, I know there is an IM that favors Bird's...but not a GM ;)...).
Neither of these are truly viable at higher levels, though. You can draw with them, but you cannot just play sharply and expect to win...in fact, you can play the game of your life with these openings and flat out lose if the opponent is strong and doesn't rattle easily.
These are probably the best gambits that stay gambits unlike queens gambit