Beginners should play open games. Does it also mean a beginner should or must play gambits?

But I feel if I "give away" a pawn, I'm already have a less advantageous position, even I know I get a fast development. But if I can't convert it to a strong attack I'm I pawn down, for nothing.

For example I consider to play the Bishops opening. On move 3 you can play d4 instead of d3 and get the Urusov Gambit with a wild tactical play. Is 3.d3 too positional for a beginner?

Though I respect him and almost always value his contributions here, I disagree with almost everything TN has had to say above.  If you read his advice, you see that it is quite good for a 2100+ player; play the Spanish; play main lines, and so on, and so forth.  But consistently, he seems to misunderstand that he's not talking to a 2150 player, but a 1500 player.  Either that, or he very mistakenly assumes that the same advice is applicable to players in both classes.

He says, "For anyone intending to reach 2700 level, it is absolutely essential to be able to play the Ruy Lopez and Open Sicilian as White and Black."  I have to yield to his apparent expertise in what it takes to make 2700, but so far as I am aware, you have to make 2200 before you make 2700.

For reasons that I have repeated innumerable times -- but which seem always to require repeating if low-rated visitors here are not to be left with bad advice -- players sub-2000 primarily need to learn (1) tactics, (2) theoretical endings (the ones with definite solutions) and (3) how to play open positions.  Again for the umpteenth time, the reason open positions are important is quite analogous to the reason that endings are important: a game of chess tends to produce them.  They are fundamental.  The object of play in very many closed or semi-closed positions is, precisely, how and under what conditions to open the game (all those ...f6 breaks in the French?  all those ...c5 breaks in the Caro? they are aimed at opening the position on favorable terms).  So if you can't play open positions well, you can't play chess well, and that is the long and the short of it.  Just the same, if you can't play king and pawn endings, what is the point of picking up a book on openings theory?  

A much stronger and more respected player than I, the esteemed Jakob Aagaard, was debating this and saying that many GM's aren't very good at play in open positions and some aren't even very good at tactics.  He may be right, but I think he means, relative to other GM's.  I respectfully maintain my view that to become a GM, an IM, an FM or even a humble USCF NM like me, you have to be able to conduct play in positions like those that arise from the Goering, the Danish, and the Blackmar-Diemer, as well as open positions of a less unbalanced kind like those arising from the Panov-Botvinnik, White's side of the QGA and Black's side of the Tarrasch.  You just have to be able to play like this because the play of any given game tends in this direction.  

So with the greatest respect, I think it is so much claptrap to tell a 1500-rated player to take up the main lines of the Spanish, which produce a semi-closed maneuvering game, play on both sides of the board, and the like, that is often way too sophisticated for players in this class.  Indeed it is claptrap to suggest that opening theory is important at all.  The point is precisely: not to study any more chess opening theory than an improving player needs to win his games.  Not to study theory! Not!!  So advice to study the main lines is a monumental pile of a proverbial substance, and it gets my freaking goat when this wretched advice is dished out and the innocent people receiving it come back and say, "Oh, thank you so much, I'm off to buy a book about the Spanish!"

These people aren't 1500-rated because they don't play the main lines of the Spanish, for crying out loud.  They're 1500-rated because they don't know how to play this game.

So please, young and developing chess players: play into open positions as much as you can, emphasize tactics and active piece play, and don't study any more theory than you need to win your games.  And don't ever make the mistake of thinking that the opening you played (assuming that it was reasonably sound) was the reason you lost any given game of chess; or that switching openings by itself will do anything for your game.

Example 2: Last week a 45/45 online-game of mine with black started: 1.d4 d5 2.Nf3 Nf6 3.e3 c5 4.c3 and I was out of my 'personal book'. 4... Qf6 5.Bd3 and Chess Assistant 10 has 13 games so far in it's data base and the highest rated players are 2500+ for both sides. After my (dubious) try 5... c4 the data base is empty and in the following moves there's no coming back to any games. The task for me is now to understand the features of the position and the plans according to this features. Studying shortly the main variation of the Colle with black may be a by-product.

Julian Hodgson was held back greatly in his career because of his reliance on the Trompowsky, an interesting but objectively inferior opening.