Benko Gambit
Give one wing pawn on move three and get two open files and a monster bishop that press White's queenside for the whole game. Your moves are the same every time (...a6, ...Bxa6, ...d6, ...g6, ...Bg7, castle) and the pressure never wears off, so a pawn down feels like nothing.
Spar this openingYour setup
- Bishopg7
- Bishopa6
- Knightf6
- Knightd7 / b6 / a6
- Queena5 / b6
- Rookb8
- Pawnsc5 d6 g6
- CastleO-O
c5 before b5. ...c5 first, and only then ...b5. The gambit works because White's pawn is stuck on d5 and the a- and b-files open onto a cramped queenside. Play ...b5 before ...c5 and cxb5 is simply a free pawn with no bind to show for it.
Ideas
The pawn pays for the a- and b-files and the g7-bishop. Squeeze, do not attack.
Most gambits want a quick attack and fall apart if it fails. The Benko wants pressure that never goes away: rooks on a8 and b8, queen on a5 or b6, bishop on g7, all pointing at a2, b2 and c3. White spends the whole game defending pawns instead of using the extra one.
When White plays e4, ...Bxf1 forces Kxf1. Their king is stuck in the centre.
Your bishop on a6 stares at f1. The moment the e-pawn moves, the bishop on f1 is undefended by anything but the king, and Kxf1 is forced. White loses the right to castle and spends two extra moves (g3, Kg2) getting the king to safety by hand.
While a knight sits on c3, ...Nxe4 just loses your knight to Nxe4.
It is the most natural grab on the board and it is wrong nearly every time in this opening. The pawn on e4 is guarded by the knight on c3. Count the defenders before you take: if c3 has a knight, leave e4 alone and play ...Bxf1 or ...g6 instead.
...g6, ...Bg7, ...O-O, ...Nbd7. Then the queen and rooks go to work.
The squeeze needs every piece, and pieces need a safe king behind them. The bishop on g7 is both a shield and a weapon, so getting it there and castling is never a wasted move. Only then bring the queen to a5 or b6 and a rook to b8.
White develops instead of taking? ...bxc4 grabs a pawn and opens the b-file anyway.
Nf3, Nd2, e3, g3 and Qc2 all leave the pawn on c4 hanging. Take it. White usually wins it back with Bxc4 or Nxc4 later, but you have the same open b-file, the same ...d6 and ...g6 setup, and no pawn deficit to worry about.
5.b6 gives the pawn back to keep the files shut. Take it with the queen and play normally.
Some players know the Benko is annoying and return the pawn immediately with b6 so that your a- and b-pawns stay blocked. ...Qxb6 gets the material back and the queen sits well on b6 anyway. Then ...d6, ...g6, ...Bg7 as usual.
Queen to a5 or b6, rook to b8, knight to b6 or c4. Aim everything at a2, b2 and c3.
Once you are castled with the bishop on g7 and knights on f6 and d7, the plan writes itself: ...Qa5 or ...Qb6, ...Rfb8, ...Nb6 heading for c4, and ...Ne8–c7 to add to the queenside. Trade pieces freely; the pressure survives every trade and even a pawn down the endgame is comfortable.
The structure
Traps
The tell: White plays e4 and your knight on f6 sees a pawn it has been trained to grab in every other opening.
The knight on c3 guards e4. After Nxe4 you have given a knight for a pawn, and you were already a pawn down for the gambit. The correct reply to e4 is ...Bxf1, taking White's castling instead of a pawn.
The tell: White supports the b5-pawn with a4 instead of taking on a6 or developing.
After ...axb5 axb5 both a-pawns are gone and the a-file is completely open. White's rook on a1 is defended only by the queen on d1, and the knight on b1 stands in the way, so ...Rxa1 simply wins a rook. White has to see this and play something other than axb5.
The tell: White plays e4 while the bishop is still on f1 and yours is on a6.
No material changes hands, but White loses the right to castle for good and will spend two more moves (g3 and Kg2) walking the king to safety by hand. Those are two moves you spend on ...g6, ...Bg7 and castling. This is the reason the bishop went to a6 rather than the knight.
Model games
Every Benko idea in thirteen moves: both pawns offered, the bishop traded for White's castling, the fianchetto, the castled king, and finally queen and rook onto the a- and b-files.
White declines and wins c4 back later. You get the same structure with level material, trade off the light bishops, and reroute the knight to c7 to support ...b5 again.
The pawn comes back on b6 and the queen is well placed there. Material is level, your development is easy, and the a-pawn is ready to march to a5 and a4.
How the coach grades a move
Every move you play gets one of six grades, from the engine’s view of the position, with the book’s verdict taking over when it is harsher. The sentence beside it is always the reason. Tap it for what the move did, why this opening wants it, and whether anything was better.
- Best
- Nothing better was available.
- Good
- Within a few percent of best. Noise at this depth.
- Playable
- A small real gap. The better move is named, play goes on.
- Inaccuracy
- Seven to twelve points of winning chances. The game stops to show you.
- Mistake
- Twelve or more, or a move the book warns against, or a broken order rule.
- Blunder
- Twenty-five or more. Usually material.
Catalogued as A57–A59 in the Encyclopaedia of Chess Openings, the reference index most databases sort by.