Many may claim that gambits are perfectly ok, and that many grandmasters all over the years have proved them to be solid enough. however... let's be honest: You don't see many gambits in the chess-elite nowadays. I believe there was way more gambit play in the early romantic-chess era. Nowadays, you will hardly ever see a grandmaster play for a King's gambit accepted in a decisive match.

Another thing that I find interesting is that I've never seen a computer play a gambit. It's probably because there's no clear advantage from it, other than maybe a couple tempi and piece development... but truth be told, if your oponent knows the line (or calculates it by force like comps) then you're prety much lost.  So it finally comes to this question: Are gambits solid at all? or will they ultimately get "solved" and dissapear from high level opening choices?

Upper-level play not having many gambits, huh? You forgot the Queen's Gambit. It's a major opening line for beginners and super-GMs alike.

The idea of the gambit is to trade material for activity/development/threats/tempo. It creates an imbalance that's harder for your opponent to calculate, usually putting them on the defensive. In the KG or the Evans, with an extremely agressive line chosen White can basically grab hold of the initiative forcing Black to respond to threats - this makes for an interesting attack or a tactical firestorm in the middlegame. In the Benko, Black trades a pawn for free piece play on the Queenside and often ownership of the dark-square diagonal, whereas White has a hard time developing because of the difficulty of pushing e4 without forfeiting casling or playing at least 5 slower preparatory moves (Nf3 g3 Bg2 0-0 Re1).

Truth be told, single pawns are not really a huge loss in the opening, especially if the compensation is good enough. Often you can regain your gambit material with interest if you can make the right threats and keep the initiative. I don't think any of the more frequently seen gambits will ever be phased out because they're what make for some of the most interesting games.

EDIT - as to "knowing the lines" - with most mainstream gambits, the person who gives up the pawn has far more choices of how to steer the game, and if they keep the initiative the opponent has to keep responding to threats. In some of the more positional gambits (once again I speak of the Benko since it's a personal favorite), I've seen 1900-rated players lose the exchange by force against it because of the active play that Black can get. There's just too much for a single person to know because of the activity that the gambiteer will get.

The Queen's Gambit isn't an actual gambit as if black takes the pawn its impossible to hold.

I know that, but it's still a gambit because White has the opportunity to turn it into one by allowing Black to safely push b5.

 I'm not sure it's impossible to hold, just requiring severely weakening moves to keep it? i don't know though, I don't play it.

Grand masters give up material all the time for initiative. And anyway why does the fact they dont play an opening have anything to do with how that opening will perform at our level ??

In some lines of the slav, where black plays c6 and white doesn't play a4, black gets to hold the pawn. Check the super-sharp botvinnik variation of the semi-slav:

The Seppuku Gambit is a very solid gambit. I have lost everytime I faced it. Here is a demonstration for those who don't know what it is:

after years of laborious study using masters notes from the past and ultra modern up to date technology, the noted chess genius Proffessor Tno Rute Ta Lla solved chess.  The result was, white wins with the Seppuku Gambit every-time.

LOL @ the Seppuku gambit.

Jester, the chess computer, will play gambits. I've seen it play the King's Gambit as well as the Queen's gambit. It is true though that computers are generally very materialistic. I've never seen a computer make a piece sacrifice unless it led to checkmate or a material gain. But then its not really a sacrifice then is it?

There are a lot of gambits played at top level these days. Two Marshall gambits (in the Ruy and in the Semi-Slav), the anti-Moscow gambit, the Poisoned Pawn Najdorf, the most fashionable line in the Queen's Indian, et cetera. Every major opening has sacrifices somewhere in its main lines, I think.

But the early gambits in 1.e4 e5 are unpopular, true.

The Perenyi Attack, Schliemann/Jaenisch and Marshall Attack are some of the other gambits played at the very highest levels.  Gambit chess is alive and well.

couldn't black retreat his bishop after capturing the queen and take his time for development? and maybe later sacrifices a piece to break white's pawns formation, being a queen ahead i think it shouldn't be that hard

I agree... and its probably just the settings!  lol.  You can program most computers to vary their openings if you want to... other wise they will almost exclusively play the popular stuff (e.g., Sicilian)

kings gambit has the reputation of suicide for white ever since Fischer essayed it, yet, as Gallagher points out, the Fischer variation is far from a refutation of the king's gambit. King's gambit didn't disappear because it was refuted, it just went out of fashion. 

Everyone have different opinions upon gambits. Korchnoi would go on and on criticising your 1.d4 Nf6 2.c4 c5 3.d5 b5, even though the Benko Gambit is no immediate disaster for black, or somewhere in the near future.

Speaking of the Benko Gambit, this is one gambit I believe that computers wouldn never understand. Unlike so many other gambits, instead of sacrificing for a shot at the enemy king, the Benko simply aim for long term queenside pressure. The key word there is longterm, something that the chess engines would never have in their vocabulary of chess terms.

I do agree with you though, it is not everyday you see GMs play a gambit, although you see Kasparov choosing the Evans Gambit before he retired once or twice. 'A pawn is a pawn', saw William Hartston, 'and anyone who throws one away willingly in the first few move ought to realize that he is commiting an offence against the natural laws of the game.'

...That's a pretty long motto to live by don't you think?

 Omicron brings up a very good point, I've never seen any chess computer play a gambit, why is that though because there's no clear advantage from a gambit?

Mind you, computers have changed a lot. Rybka likes sacrifices, perhaps a bit too much. My club's trainer (an IM) says that when you're analyzing a pawn sac with Rybka, and it says it's only +0.1 for you, then you should really start doubting whether there is enough compensation.

You must not be looking very hard.  Computers these days are capable of making purely positional sacrifices.  There's at least one gambit line named after Fritz.

A couple of tempi and a lead in development sounds like a recipe for a won game.