What Is a Poker Solver? Equilibrium and Exploitative Solvers
What a poker solver computes, how it works, the two kinds of solver and what a solver does not do. We make GTOkiller, an exploitative solver, so we say so where it comes up; judge it by the same criteria as the rest.
A poker solver is software that computes the strategy for a poker spot by repeatedly improving both players’ strategies until neither can gain by changing. An equilibrium (GTO) solver gives the strategy against a perfect opponent; an exploitative solver such as GTOkiller fixes the opponent to how a real player pool measurably plays and computes the strategy that wins most against it.
What a poker solver computes
A solver takes a spot (the positions, the stack depth, the range each player arrives with, the board and the bet sizes allowed) and returns a strategy for every hand of each range at every decision: how often to check, bet each size, call, raise or fold, and the EV of each action.
All the decisions of a spot form the game tree, and each decision point is a node. A spot played from the flop with a few bet sizes per street already has thousands of nodes once the turn and river cards are counted; GTOkiller’s full trees have about 500,000. That is why solvers run on computers and why their output is studied, not memorised.
How a solver works: CFR and Nash equilibrium
Most poker solvers use counterfactual regret minimisation (CFR) or a variant of it. The algorithm starts both players with arbitrary strategies and goes through the tree over and over. At each node it measures how much each player regrets not having taken each action and moves the strategy toward the actions with the most regret. After enough iterations, the average strategies converge.
In a two-player spot they converge to a Nash equilibrium: a pair of strategies in which neither player can win more by changing theirs. That is what GTO (game theory optimal) means in poker.
No solver reaches the exact equilibrium. It stops when the exploitability, the most a perfect opponent could win against the strategy, drops below a target that is usually a small fraction of the pot.
Two kinds of solver: equilibrium and exploitative
An equilibrium solver answers "how do I play so that nobody can exploit me?". Desktop programs such as PioSolver and GTO+, cloud solvers such as Deepsolver and presolved libraries such as GTO Wizard work this way: the opponent is assumed to play perfectly, so the answer is the GTO strategy. GTOkiller’s guide to the best poker solvers in 2026 compares them by job.
An exploitative solver answers a different question: "given how this pool actually plays, what wins the most?". Instead of letting the solver choose the opponent’s strategy, it fixes the opponent’s frequencies to measured population data and computes the best response to them. We make GTOkiller, an exploitative solver, so read this as ours: GTOkiller measures how a real player pool plays in every node of the tree and computes the highest-EV strategy against its tendencies and leaks, keeping a balanced base where a deviation would gain little.
Against a real pool, that strategy has more EV than GTO, because GTO does not go after the pool’s mistakes. Is there anything better than GTO? puts numbers on four of those mistakes, with figures from GTOkiller’s Population Report 2026.
Equilibrium and exploitative solvers, side by side
| Equilibrium (GTO) solver | Exploitative solver | |
|---|---|---|
| The question it answers | How to play so nobody can exploit you | How to win the most against how a pool really plays |
| Opponent model | A perfect opponent, computed by the solver | The measured frequencies of a real player pool |
| Output | The equilibrium (GTO) strategy | The highest-EV strategy against the pool, next to GTO |
| Strength | It cannot be exploited; the base of modern study | It punishes the mistakes the pool actually makes |
| Limit | It does not go after the opponents’ mistakes | It is only as good as the data, and an opponent who adapts can exploit it back |
| Examples | PioSolver, GTO+, Deepsolver, GTO Wizard | GTOkiller |
The question it answers
- Equilibrium (GTO) solver
- How to play so nobody can exploit you
- Exploitative solver
- How to win the most against how a pool really plays
Opponent model
- Equilibrium (GTO) solver
- A perfect opponent, computed by the solver
- Exploitative solver
- The measured frequencies of a real player pool
Output
- Equilibrium (GTO) solver
- The equilibrium (GTO) strategy
- Exploitative solver
- The highest-EV strategy against the pool, next to GTO
Strength
- Equilibrium (GTO) solver
- It cannot be exploited; the base of modern study
- Exploitative solver
- It punishes the mistakes the pool actually makes
Limit
- Equilibrium (GTO) solver
- It does not go after the opponents’ mistakes
- Exploitative solver
- It is only as good as the data, and an opponent who adapts can exploit it back
Examples
- Equilibrium (GTO) solver
- PioSolver, GTO+, Deepsolver, GTO Wizard
- Exploitative solver
- GTOkiller
Node locking: exploiting with a GTO solver
Equilibrium solvers can also exploit, through node locking: you fix the opponent’s strategy in one node, for example a big blind that folds 60% to a c-bet, and the solver computes the best response. It is the classic way to study exploits, and it works well against one opponent you know.
Against a whole pool it has two limits: the frequency you lock is your guess, not a measurement, and every node you do not lock is still solved as perfect play. One example from GTOkiller’s population data: on K♠8♥4♦, small blind against big blind, the pool folds 35% to a one-third pot c-bet, where GTO folds 23%. The guide on node locking vs population data goes through both limits.
What a solver does not do
A solver is a study tool, with clear limits:
- It does not read your opponents: its answer is only as good as the ranges, sizes and opponent model it is given
- It does not play for you: using a solver or any real-time assistance during a hand breaks the rules of online poker sites, so solvers are for study away from the table
- It does not explain itself: it returns frequencies and EV, not the reasoning, which is why study tools add explanations such as GTOkiller’s AI Coach
- It does not cover every format equally: multiway pots, tournaments with ICM and PLO need specific solvers or simplifications
An exploitative solver
GTOkiller: Exploiting your player poolThis is us
An exploitative poker solver. Instead of assuming opponents play GTO, it measures how a real player pool plays and computes the highest-EV strategy against its tendencies and leaks across the whole game tree.
- Strengths
- Strategies by stake and player profile, valid on any site; three strategies side by side (exploitative, GTO and population); an AI Coach that names the leak each line exploits and its EV.
- Limits
- 6-max NLHE cash only, no custom spots, and the solver needs a screen at least 1024 pixels wide.
- Price
- Free plan with one board a day; Edge €39 / $49 and Killer €79 / $99 a month.
Which kind of solver to use
| If you want to | Use |
|---|---|
| Learn balanced play and how not to be exploited | An equilibrium solver or a presolved GTO library |
| Study a custom spot with your own ranges, stacks and sizes | An equilibrium solver such as PioSolver, GTO+ or Deepsolver |
| Exploit one opponent you have read yourself | Node locking in an equilibrium solver |
| Win the most against how the pool at your stake plays | An exploitative solver such as GTOkiller |
Learn balanced play and how not to be exploited
- Use
- An equilibrium solver or a presolved GTO library
Study a custom spot with your own ranges, stacks and sizes
- Use
- An equilibrium solver such as PioSolver, GTO+ or Deepsolver
Exploit one opponent you have read yourself
- Use
- Node locking in an equilibrium solver
Win the most against how the pool at your stake plays
- Use
- An exploitative solver such as GTOkiller
When GTOkiller is not the answer
- Tournaments, PLO or ICM: GTOkiller covers 6-max NLHE cash only
- Custom spots with your own ranges, stacks or bet sizes
- A stake that is not covered yet; new ones are added by demand
- Studying only on a phone: the solver needs a screen at least 1024 pixels wide
Frequently asked questions
- What is a poker solver?
- Software that computes the strategy for a poker spot: for every hand in each player’s range, how often to check, bet, call, raise or fold, and the EV of each action. Most solvers compute the GTO strategy; an exploitative solver such as GTOkiller computes the strategy that wins most against how a real player pool plays.
- How do poker solvers work?
- Most use counterfactual regret minimisation (CFR): they go through the game tree over and over, moving each player’s strategy toward the actions they regret not taking, until both strategies converge on a Nash equilibrium within a small exploitability.
- What is an exploitative poker solver?
- A solver whose opponent is not assumed to play perfectly: it takes the measured frequencies of a real player pool as the opponent model and computes the highest-EV strategy against them. GTOkiller is an exploitative poker solver for 6-max NLHE cash, with strategies by stake and player profile.
- What is the difference between a GTO solver and an exploitative solver?
- A GTO solver computes the strategy nobody can exploit, assuming a perfect opponent. An exploitative solver computes the strategy that wins most against a given opponent model; in GTOkiller, the measured play of a player pool. The first is the safe baseline; the second is where the extra EV against real pools comes from.
- Which solver uses real population data?
- GTOkiller: it is built on population data by stake and player profile and computes the strategy against it across the whole tree. MDA platforms show population data and mostly leave the strategy to you, and GTO solvers only use it if you type the frequencies in with node locking.
- Is using a poker solver cheating?
- Studying with a solver away from the table is allowed, and it is how most serious players learn. Using a solver or any real-time assistance while you play breaks the rules of online poker sites. GTOkiller is a study tool: you open a spot, study the strategy and then play on your own.
- Do I need a solver to win at poker?
- No, but it speeds up learning. A presolved library or a free solver is enough to start with GTO; an exploitative solver helps once you want to profit from how your pool actually plays. The GTOkiller free plan gives one board a day with the full tree.
Try the exploitative solver for free
One board per day with the full decision tree, the three strategies and the AI Coach.