Benko Gambit: how to play the queenside pawn sacrifice, and what the engine charges for it

Short answer

The Benko Gambit gives up a pawn on move three to open the a- and b-files and press on the queenside for the rest of the game. This guide walks through the main line move by move — the bishop trade that forces the white king to walk to g2 — defines the difference between tactical compensation and permanent positional compensation, and reports ten positions measured on Stockfish 18 at depth 26: the accepted line, the six common declines, and a comparison with the Modern Benoni, which reaches similar numbers without sacrificing anything.

The essentials in a few lines

• The Benko does not equalise by engine standards. After 4.cxb5 a6 5.bxa6 Bxa6, Stockfish 18 gives +0.83 at depth 26, and in the typical move-12 position the evaluation reaches +0.95. Anyone playing the gambit should know they are playing a position the engine considers worse.
• The compensation is positional and permanent rather than tactical: the open a- and b-files and the backward white pawns on a2 and b2 are structural features that survive every exchange. That is the opposite of an initiative gambit, whose compensation expires within ten moves.
• The main line makes the white king walk. After 6.Nc3 d6 7.e4 Bxf1 8.Kxf1 g6 9.g3 Bg7 10.Kg2, White has the king on g2 and the rooks disconnected — the price of holding the extra pawn.
• The most annoying decline is not the most aggressive one. Against 5.f3 Black reaches +0.03, technical equality; against 5.b6, +0.20 after 5...e6. The lines where White hands the pawn back are the ones that erase the Benko's typical play, not the ones that try to hold on to it.
• The comparison that sizes up the cost: the Modern Benoni, with a similar structure and no sacrifice at all, measures +0.60 under the same conditions. The Benko pawn buys a plan and two files — it does not buy back half a pawn of evaluation.

What the b5 pawn buys

The idea is easy to state and hard to believe the first time: Black gives up a pawn on move three, gets no immediate threat in return, and still plays the following game with the plan already written. What Black receives is two open files pointing at the wing where the white king almost never goes, and a structure where the black pieces have better squares than the white ones — the a8 rook is already on the right file before the file even opens.

The name comes from the Hungarian-American grandmaster Pal Benkő, who systematised the gambit and took it to elite level; in the Russian tradition the same opening is called the Volga Gambit, and both names describe the same sequence of moves. The ECO codes run from A57 to A59, and the split between those three codes is precisely between accepting fully, accepting halfway and declining — which already tells you that White's decision on move four is the opening's main subject.

For repertoire purposes, the Benko solves a specific problem: giving Black a game with initiative against 1.d4 without depending on long theory. In the classical closed defences Black usually equalises by defending; here Black attacks from the start, and the price is a pawn that rarely comes back. It is a trade that suits players who lose games through passivity — and does not suit players who already struggle to judge when an attack is over.

Tactical compensation and positional compensation

The distinction matters because it changes how the position is played. In an initiative gambit, any move that does not increase the pressure is a move wasted, and the practical rule is to play fast and forcing. In the Benko it is the reverse: Black can trade queens, can simplify, can take the game to move 40 — the a- and b-files stay open, and the white pawns on a2 and b2 still need pieces to defend them. No clock is running against the compensation.

That is also why the Benko is one of the few gambits you are happy to take into an endgame. In a position with few pieces a pawn down is usually decisive; here the black rook landing on a2 or b2 either regains material or ties the white pieces to defence. The gambit is not betting on mate: it is betting that the resulting structure is easier to play from the black side for forty moves.

That permanence was tested for this guide, and the test is easy to repeat: let Stockfish 18 play both sides from the move-14 position, evaluated at +0.84. Fourteen moves later the queens were off the board, the black rook sat on the second rank at c2, and the evaluation had dropped to +0.29 — simplification, which normally favours the side with more material, favoured here the side with the missing pawn. What the test does not say is that the compensation is worth a whole pawn: the measurements in the next section show what the engine charges, and the number is not zero. The argument for the Benko is not that it equalises, but that the position it produces is easier to play than the evaluation suggests.

• Tactical compensation: initiative, exposed king, development — expires within a few moves.
• Permanent positional compensation: open files, backward pawns, weak squares — survives exchanges.
• The Benko is the second kind, which is why trading queens is no problem for Black.
• Permanent does not mean sufficient: the engine still charges around half a pawn.

The main line, and the white king on foot

The sequence has a two-way logic. White wants to keep the pawn and occupy the centre with e2-e4; to play e4, the f1 bishop must be left unguarded, and that is when the trade on f1 happens. Black ends up without the bishop pair and White without castling — neither side has a better option, and the resulting position is what defines the modern Benko.

After 10...O-O 11.Nf3 Nbd7, Stockfish 18 measures +0.95 at depth 26, with 12.a4 as the best try. It is the peak evaluation of the whole opening: White still has the pawn, has already organised the king and has not yet felt the pressure on the files. If there is a moment when the Benko looks like a mistake, this is it — and it is exactly the moment when Black starts playing the plan.

Black's plan is always the same and fits in three moves: queen to a5 or b6, rooks to a8 and b8, the d7 knight to b6 or c5 depending on whether White pushes a4. There are no variations to memorise after move 12 — there is a piece set-up to assemble. In the positions measured for this guide, once that set-up is complete the evaluation drops: after 12.a4 Qa5 13.Ra3 Rfb8 the engine gives +0.72, and its preferred continuation for White becomes 14.h4, a flank move with nothing to do with the extra pawn.

What the engine charges for the pawn

It is worth facing the number rather than working around it, because it is the argument of everyone who advises against the gambit. On the engine's scale a pawn is worth about 1.00. If the accepted Benko measured +1.00, the compensation would be zero; if it measured 0.00, it would be complete. It measures +0.83 to +0.95, which means partial compensation — somewhere between a third and half of the pawn comes back as activity, and the rest is the gap you play out on the board.

The comparison with the Modern Benoni is useful because both produce positions from the same family: black pawn on c5, white pawn on d5, bishop on g7, play on the half-open file. In the Benoni, measured at the same depth, the engine gives +0.60 after 8.h3 O-O — and White's second and third choices there are worth less, between +0.41 and +0.50, which suggests the edge depends on finding one specific move. In cold numbers, then, the Benoni is the cheap version of the same idea, and that is worth saying to anyone choosing a repertoire.

What the numbers do not measure is what sustains the gambit in practice, and it pays to be specific rather than vague: in the Benko, Black has a plan that requires no calculation — double the rooks, press a2 and b2 — while White has to decide, move by move, how to defend two backward pawns without loosening the centre. Asymmetric positions where one side has a fixed plan and the other has decisions cost the second player clock time, and that is where the half pawn comes back.

The declines, and which one actually hurts

There are two families of decline. The first holds the pawn some other way or shores up the centre before committing — 4.Nf3 (+0.44), 4.Nd2 (+0.27), 4.a4 (+0.29), 4.Qc2 (+0.18) — and none of them produces an advantage: they are moves that swap the material argument for an ordinary positional game in which Black is close to equal and without the usual target.

The second family returns the pawn deliberately, and it is the one that demands more from Black. Against 5.b6 the right move is 5...e6, and the engine drops to +0.20: Black recovers the b6 pawn later and the game turns into a Benoni with a normal structure. Against 5.e3 the number rises to +0.58, the highest among the declines — White finishes development before the file opens, and it is the line that most resembles a practical refutation of Black's plan.

The most surprising result in the measurements is 5.f3, the Zaitsev set-up, which has a reputation for aggression: it leaves Black at +0.03 after 5...axb5 6.e4 Qa5+, that is, technical equality. The inversion is worth recording, because it contradicts the intuition of anyone meeting the line for the first time: the positions where White works hardest to keep the pawn are the ones where Black ends up best. Benko players should fear the calm return, not the attempt to hold on.

Who the Benko suits, and how to train it

Training the Benko saves the resource most anti-1.d4 repertoires spend: memory. There are no twenty variations to learn, only three positions to recognise — accepted with e4 and the king on g2, accepted with the fianchetto, and the family of returns. The part that genuinely requires study is the endgame, and that is the inversion that tends to surprise people: gambit players study attack, and the Benko asks for endgame technique with rooks on an open file.

There is an honest test for anyone considering the opening: review your own games against 1.d4 and count in how many the problem was having no plan, and in how many it was losing material. In the first case the Benko fixes what needs fixing. In the second it adds a difficulty — playing a pawn down for thirty moves — on top of an existing one, and the Modern Benoni does the same structural work without the bill.

On faster time controls, practice and measurement point the same way: the Benko yields more in long games, because the value of the compensation appears with a whole game ahead and Black's set-up almost plays itself. In blitz the missing pawn weighs more than the plan, since White's defence — imprecise in the early moves — is the part the opponent solves with time, and on a short clock nobody has time to prove anything on the b-file.

Frequently asked questions

Openings mentioned in this guide

• Benko Gambit (A57)
• Modern Benoni Defense (A60)
• King's Indian Defense (E60)
• Budapest Gambit (A51)