Benko Gambit eventually always accepted

I am a player of the Benko Gambit and what I've noticed is that they'll often decline the gambit, and that disturbed area of him attacking my undefended pawn remains.  We'll be developing like normal, fianchettoing my king bishop, playing d6, playing Nd7, etc. but in the end they'll ALWAYS accept my gambit.  They also play a mistake.  They also trade of light squared bishops.  It's a mistake because my light squared bishop is not as important as my dark squared bishop and it's the opposite for him.  So I get an even bigger advantage...  Then I start harassing them on the queen side as with any Benko Gambit position. 

So my question is, why is it that they will always accept the gambit even though it may not be right off the bat?

post a game?

In response to the OP, accepting the gambit is the strongest, and the bust to the Benko.

Sure, some players will play 5.b6, just to try to plug up the Queenside, but White's strongest move is to take the pawn, and then there are two lines where White gets a big advantage, one of which even Kasparov (not Garry, the other one) acknowledges in his book from a year ago as being a problem for Black in his addendum at the end, and those two lines are the Fianchetto Variation (which even Kasparov agrees) and the King Walk variation, where the LSB's are traded, but not at White's accord, at Black's, where Black takes on f1.  White taking on a6 only accelerates the play for Black.

I still to this day will never play the Benko Gambit again as Black, and will never advocate the Benko Gambit to anybody unless they want to lose.

The same can be said, by the way, about the Alekhine (Black) and King's Gambit (White).  Don't play them!  You'll simply get creamed every time once you start playing against players that really know what they are doing!  I've played my fair share of games against players 2200 to 2650, and am well aware of what can and can't be played against these guys if you want any chance at all!

5.b6 is a very annoying move to meet. Black either has to suffer a positional squeeze, or liquidate quickly to a rougly level middlegame by 5...e6!, where his winning chances are minimal, if existing at all.

The fianchetto/ Epishin is not much of a threat anymore. Black has three different ways to get a good game, Sergey Kasparov's ...Nfd7 not being one of them.

The king walk variation combined with 12.a4(!) is certainly more threatening. Here Black has an interesting, but quite risky way toi generate play- it involves winning a piece for three pawns, but while this particular continuation looks OK for correspondence play, it's quite impractical for OTB play, since Black's pawn structure becomes a wreck.

There is also a new, interesting approach advocated by GM LuckyTiger (AKA Valery Aveskulov) in his recent Benko book revision, which looks OK, but so far there is little evidence about it.

Finally there is another very new approach by white, namely 1.d4 Nf6 2.c4 c5 3.d5 b5 4.cb5 a6 5.ba6 g6 6.a7!? which has been met only in hi-level centaur chess, and so far scoring great, but it is probably not so dangerous.