Free GTO poker solver
See how often to bet, check or fold in heads-up postflop spots. Review your hands, explore precomputed flop strategies and practice with EV-loss feedback. Custom solves run on your own device. Nothing to install, nothing to pay.
Open the solver →Open Study Spots →
New to solvers? Start with Study Spots—no setup needed.
Enlarge screenshot
A GTO solver takes two ranges, a board, the pot, the stacks and the bet sizes you allow, then searches for a strategy neither player can improve on alone. The output is not one move but a frequency for every hand: how often to bet, check, call, raise or fold. GTOZero is an online poker solver that runs that search inside your browser, so you can test the exact spot you played instead of the closest chart.
A chart works differently. It is an answer someone worked out earlier for a typical spot; a solver works the answer out again for the ranges, board and sizes you give it. Change the flop and the strategy changes with it.
You will see the same kind of program called a poker solver, a GTO poker solver or a postflop solver. They describe the same thing — software that computes equilibrium frequencies for a spot — and GTOZero is all three.
GTO stands for Game Theory Optimal: a pair of strategies in which neither player can improve their expected result by changing their own play alone. In practice that means balanced, mixed frequencies — a hand is bet some of the time and checked the rest — so an opponent cannot profit from reading a pattern.
That makes GTO a reference point for sound play: the strategy you measure your own decisions, and an opponent's mistakes, against.
The part that surprises most players is mixing. At equilibrium the same hand is often played more than one way — bet some of the time, check the rest — so nobody can read your hand from your action. That is why a solver answers with percentages rather than a single instruction, and why grading a decision is about how much value it gives up, not about matching one “correct” move.
Enter both ranges, the flop (and a turn or river if you want a fixed runout), the starting pot, the effective stack and bet and raise sizes for each street. The calculation runs on your computer's CPU through WebAssembly — nothing is sent to a server to be solved. It stops at the target exploitability you set (0.3% of the pot by default) or at the iteration limit.
Thirteen worked examples open instantly. Press ⚡ View results and the flop strategy appears straight away, so your first session can be about reading frequencies, equity and EV rather than settings. To change the ranges or follow the hand to the turn and river, Solve it yourself loads the same inputs into the solver.
Pick the situation (Open · Call, 3-bet pot or 4-bet pot), the seats, your two cards and the flop, then enter the actions in the order they happened. Analyzed decisions come back as Best play, Acceptable or Review this play, with a Total loss in big blinds; decisions the hand very rarely reaches are shown for reference only and left out of the total. Flop decisions use precomputed results; turn and river decisions are calculated on your device. Ranges come from the app's preflop charts, so there is nothing to type. The basis is online 6-max cash with 100bb stacks, heads-up pots only, and you enter each hand yourself, one hand at a time.
Enlarge screenshot For the main seat combinations of single-raised, 3-bet and 4-bet pots, all 1,755 distinct flops (suit-swapped flops counted once) have been solved in advance to a 0.3% target. Flop study shows one flop: action frequencies and a 13×13 strategy grid, and you can follow each action to the end of the flop. Flop reports averages all 1,755 flops, weighted by how often each one is dealt, grouped by high card, suits, pairing or connectedness. In these pots flop bets are fixed at 33% and 75% of the pot with 60% raises; the button-open, big-blind-call data starts from 100bb stacks with no rake and 60% turn and river bets. Turn and river play is not stored — Open in solver calculates it from there. A small blind vs. big blind limped pot is also available.
Enlarge screenshot 
Enlarge screenshot The trainer deals hands from solved spots in proportion to how often you would really hold them, asks for your action, and grades it by EV loss relative to the pot: Best play, Acceptable or Review this spot. A Daily Puzzle, a Review queue, Leak Finder and an optional RNG mode for mixed strategies are built in. From any result screen, Practice this spot turns your own solve into trainer questions.
Enlarge screenshot On the results screen, Lock this node’s strategy fixes how one player acts at that point. Re-solve and compare before and after to see how the other player's strategy should shift when an opponent makes a mistake — for example, folding too often to a river bet.
Open a Study Spot first, so you know what a finished result looks like. Then build your own spot through the five tabs on the left, in order:
Chips are entered as whole numbers; the app suggests 10 chips = 1bb, so a 55-chip pot is 5.5bb. If a tree needs more memory than the browser allows (about 3.9GB), switch Precision mode to 16-bit integer or simplify the tree.
Typing two ranges from nothing is the slowest part of a first solve. These are the ranges behind the single-raised-pot Study Spots — button opens, big blind calls. Paste them into ① and ②, pick a flop and keep the default sizes as a starting point for your first solve.
TT-22,AJs-A2s,KJs-K2s,QJs-Q2s,J4s+,T6s+,96s+,85s+,75s+,64s+,54s,AJo-A2o,K9o+,Q9o+,J9o+,T8o+,98o22+,A2s+,K5s+,Q6s+,J7s+,T7s+,97s+,86s+,75s+,64s+,54s,A2o+,K9o+,Q9o+,J9o+,T8o+,98oCheck how these two ranges match up on any flop with the free poker equity calculator →
Start with the range summary on the right, then look up individual hands in the grid on the left. Use the action line at the top to move between decisions in the hand.
| Area | What it shows | How to use it |
|---|---|---|
| Top strip | The action line — flop, bet, call, turn and so on | Click a step to see the strategy at that point in the hand |
| 13×13 grid (left) | One cell per hand, split by color into action frequencies | Green marks checks and calls; red marks bets and raises, with darker shades for larger sizes. Hands outside the range appear dimmed |
| Action tiles (top right) | Each action's frequency and combo count for the whole range | The quickest answer to "how often does this range bet here?" |
| Hand categories (middle right) | How the range hits the board: top pair, draws, nothing | Shows which player the board helps more |
| Table (bottom right) | Weight, equity, EV, EQR and action frequencies per hand | Sort by any column; export the summary to CSV |
Try it on the first Study Spot, a dry A♥7♦2♣ flop with the big blind to act. The tiles read check 98.2% (455.5 combos), bet 1.8bb 1.0% and bet 4.1bb 0.9%. The All row of the table puts the range at 464.0 combos with 45.1% equity, 2.09bb of EV and 84.0% EQR.
That last number is worth a pause. EQR — equity realization — compares a range's EV with its equity multiplied by the pot. At 84%, the big blind's EV here is about 84% of that equity-based reference value. Position, the two ranges and the board all feed into the figure, and out-of-position ranges often realize less than their raw equity.
Yes. There is no paid tier and no limit on how many spots you solve. Here is what is included, and its limits.
| Feature | Included | Details |
|---|---|---|
| Solve flop, turn and river | Yes | Full heads-up postflop tree |
| Solve preflop | No | The app's preflop charts are reference ranges from public material, not solver output |
| Custom bet and raise sizes | Yes | % of pot, multiples, all-in and geometric sizes |
| Separate donk (lead) sizes | Yes | Give the out-of-position player its own leading sizes |
| Rake and rake cap | Yes | Set them to match your game |
| Edit the tree node by node | Yes | Add or remove actions at one point in the hand |
| Node locking and re-solve | Yes | Fix one player's frequencies at a node and compare before and after |
| Precision modes | Yes | 32-bit float or 16-bit integer, to fit larger trees in browser memory |
| Target exploitability | Yes | 0.3% of the pot by default; lower is more precise and slower |
| Save, import and export ranges and settings | Yes | Keep your own ranges and bet sizes for reuse |
| Export results to CSV | Yes | Open the summary in a spreadsheet |
| Share a spot by link | Yes | Send the exact spot to a study partner |
| Study Spots | Yes | 13 examples with instant results |
| Flop study | Yes | Precomputed flop strategies for the main seat matchups · flop bets 33% / 75%, raises 60% |
| Flop reports | Yes | Average strategy over 1,755 flops, weighted by how often each is dealt, grouped by board type |
| Hand review | Yes | Grade each decision of a hand you played · 6-max cash, 100bb, heads-up pots |
| GTO Trainer | Yes | Graded by EV loss · Daily Puzzle · Review queue · Leak Finder |
| RNG practice | Yes | A 1–100 number per hand for mixed strategies · off by default |
| Practice your own solves | Yes | Turn a result into trainer questions, stored on this device |
| Limped pot (small blind limps, big blind checks) | Yes | Flop study and Flop reports for the limped pot |
| Multiway pots | No | Two players only |
Yes. GTOZero includes a free GTO poker trainer in the same app. It deals you a hand at a real decision point from a solved spot and you choose an action. With RNG mode off, it grades the gap between your action's EV and the best action's EV, measured against the pot at the start of the spot.
Why a share of the pot? Take an action that loses 0.08bb of EV. With a 5.5bb starting pot that is 1.45%, so it is graded Review this spot; with a 22.5bb starting pot it is 0.36%, so it is Acceptable. For those two pots the limits work out to 0.02bb and about 0.055bb (single-raised pot) and about 0.079bb and 0.225bb (3-bet pot). The minimum thresholds keep very small EV differences from being graded too harshly.
The trainer samples each solved range by its hand weights, so common holdings come up often and rare ones rarely. Questions come from many decision points across the solved spots — more than ten thousand combinations — and each pot type — single-raised, 3-bet or blind-versus-blind — can be practiced separately. Your recorded answers feed three ways of tracking progress: streaks, a breakdown of weak spots, and a Review queue for revisiting the decisions that cost EV.
Open the Study Spots list in the app, choose an example, then press ⚡ View results to see its flop strategy without running a solve. Ranges follow standard 100bb online play. For a controlled experiment, load a spot into the solver, change one input such as the board or a range, and compare the new strategy with the original. This helps you see how one changed input affects the strategy.
Out of position: big blind (caller) · In position: button (opener) · Pot 5.5bb · Stacks 97.5bb
Out of position: big blind (3-bettor) · In position: button (caller) · Pot 22.5bb · Stacks 89bb · SPR about 4
Out of position: small blind (opener) · In position: big blind (caller) · Pot 6bb · Stacks 97bb
GTO Wizard and PioSOLVER are the two tools players most often compare a solver with, and each was built for a different job. Both cover ground GTOZero does not — preflop solving, for one. If you are looking for a GTO Wizard alternative, the table shows which jobs GTOZero can take on and which it cannot. Each cell is based on the tool's own official pages; prices are left out, so check those on each site.
| GTOZero | GTO Wizard | PioSOLVER | |
|---|---|---|---|
| Where the calculation runs | In your browser, on your own device | Online, on GTO Wizard's side — no solving hardware needed on yours | Windows desktop application (Mac through Parallels or Boot Camp); cloud solving also available through PioCloud |
| Heads-up postflop solving | Yes | Yes | Yes |
| Preflop solving | No — reference charts only | Yes — including multiway preflop with up to 9 players | Yes — heads-up preflop in the Edge version (at least 64GB of RAM and a fast modern CPU) |
| 3-way postflop pots | No | Yes | — (described as a heads-up solver) |
| Tournaments and ICM | No ICM — custom heads-up postflop spots can be entered; the built-in data uses cash-game ranges | Yes — ICM, with payout structures and bounties | Yes — ICM calculations |
| Ready-made spots to browse | Yes — 13 Study Spots and 1,755 precomputed flops per main seat matchup | Yes — libraries from preflop to river | — |
| Node locking | Yes | Yes | Yes |
| Aggregated flop reports | Yes — 1,755 flops | Yes — 1,755 flops | Yes |
| Trainer | Yes — graded by EV loss | Yes — instant GTO/EV feedback | Yes — built-in trainer |
| Review your own hands | One hand at a time, entered by hand (6-max cash, 100bb, heads-up pots) | Yes — upload hand histories | Set up and solve the spot yourself; hand-history upload not stated |
| Scripting | No | — | Yes |
| Source code | Open source (AGPL-3.0) | — | — |
— = not stated on the official pages we checked (October 2026). Sources: gtowizard.com · GTO Wizard blog · PioSOLVER docs · PioSOLVER products · PioCloud · TexasSolver
So which should you use? GTO Wizard covers multiway solving, ICM study and uploaded hand histories. PioSOLVER offers heads-up preflop solving in its Edge version, ICM calculations and scripting. GTOZero does none of those; it focuses on heads-up postflop study — solving your own spot, reading precomputed flops and drilling — free, with nothing to install, on whatever device you have. Many players use more than one tool, and running the same spot in two of them is a good way to check your inputs.
TexasSolver is another free solver: an open-source (AGPL-3.0) solver for Texas Hold'em and short deck, available to download for Windows, macOS and Linux and free for personal use. GTOZero runs in a browser with no separate installation, and includes Study Spots, hand review, precomputed flops and a trainer.
GTOZero continues development of the WASM Postflop engine (AGPL-3.0), whose original website now carries a notice that updates have ended. On top of the engine it adds Study Spots, hand review, precomputed flop data and the trainer.
Four short walkthroughs recorded on the same app.
Yes. Custom solves, Study Spots, hand review, Flop study, Flop reports and the GTO Trainer are all free. There is no paid tier.
No. The solver is a web app built on WebAssembly and opens in Chrome, Edge, Firefox or Safari. You can add it to your home screen if you want it full screen, but that is optional.
On your own device. Your browser downloads the solver engine and uses your CPU, so a faster computer solves faster. Precomputed flops for hand review and Flop study are downloaded only for the flop you open.
Accuracy is measured as exploitability: how much an opponent could gain against the strategy, as a percentage of the pot. Custom solves default to a 0.3% target; check the final exploitability, because a solve can also stop at the iteration limit. The hand review and Flop study data uses a 0.3% target, and the Study Spots and trainer use 0.5%. Results are only as good as the ranges and bet sizes you give the solver.
No. It solves heads-up postflop spots. For preflop the app has reference charts, which are consensus ranges from public material rather than solver output.
Yes. Study Spots, hand review, Flop study and the GTO Trainer all run on phones. Custom solves are best done on a desktop browser, especially on iPhone, where Safari limits the solver to one thread.
A program that computes game theory optimal strategy for a spot. You give it both players' ranges, the board, the pot, the stacks and the bet sizes allowed, and it works toward an equilibrium, returning how often each hand should bet, check, call, raise or fold. Unlike a chart, it recalculates for the exact spot you enter.
It builds a game tree from the two ranges, the board, the stacks and the bet sizes you allow, then plays that tree against itself over and over. Each pass nudges every hand's action frequencies toward whatever earns more, until neither player can gain much by changing — the remaining gap is the exploitability, and the solve stops when it reaches your target or the iteration limit. GTOZero runs this iterative process in WebAssembly on your own CPU.
A range is the set of hands a player can hold in a spot. GTO ranges add how often each hand takes each action at equilibrium — preflop that means which hands open, call or 3-bet from each seat; postflop the solver splits the range into the hands that bet and the hands that check. In GTOZero you enter both ranges on the 13×13 grid or paste the starter ranges on this page. The app's preflop charts are reference ranges built from public material, not solver output.
Study a spot, then drill it. Open a Study Spot or a board in Flop study, predict the strategy before you look, and compare. Then practice the same kind of spot in the GTO Trainer and revisit the decisions that cost EV in the Review queue. Hand review closes the loop with hands you actually played.
In a heads-up model without rake, an exact equilibrium strategy protects its expected value against any opponent: no opposing strategy can do better against it. That guarantee does not automatically carry over to raked or multiway games, and solvers only approximate equilibrium within the ranges and bet sizes you enter — exploitability measures how close they get. GTO can gain from an opponent's mistakes, but not from every deviation, and it is not a promise of profit after rake. Against a player with clear leaks, a strategy adjusted to those leaks can win more; node locking lets you study that adjustment.
Open a Study Spot before you set anything up — the results are already there, so you can learn to read them first. Then build your own spot through the five tabs in order: ① OOP Range, ② IP Range, ③ Board, ④ Bet Sizes, ⑤ Run Solver. Keep the default bet sizes the first time, and use the starter ranges on this page if you don't want to type ranges from scratch.
Yes. Flop study shows precomputed flop strategies straight away: choose Open · Call, 3-bet pot or 4-bet pot, the seats and a flop, and you get the first player's action frequencies and a 13×13 grid, and can follow each action to the end of the flop. The basis is 6-max 100bb, solved to a 0.3% target, with flop bets of 33% and 75% and raises of 60%. Turn and river are not stored — Open in solver calculates them, or other sizes, on your device.
Flop reports answers that. It averages strategy over all 1,755 distinct flops, weighted by how often each is dealt, for the first player's decision and for the in-position player after a check. Choose a grouping — High card, Suits, Pairing or Connectedness — to compare betting across board types; each row of the table links straight to that board in Flop study.
Turn on RNG in the GTO Trainer. Each hand then comes with a number from 1 to 100, and the 1–100 scale is divided among the actions in proportion to how often the solver takes each one with that hand. Taking the action whose range contains your number counts as Best play. It is off by default, and it is not used in the Daily Puzzle.
GTO Wizard covers more ground: preflop solving (including multiway preflop), 3-way postflop pots, tournaments with ICM, large solution libraries and hand history uploads. GTOZero covers none of those. What it does is heads-up postflop solving that runs in your own browser, with Study Spots, precomputed flops, hand review and an EV-loss trainer — all free. The comparison table above lists each point with its source.
PioSOLVER is a Windows desktop application, with PioCloud for solving in the cloud and accessing results from anywhere. Its Edge version adds heads-up preflop solving, and it supports ICM calculations and scripting. GTOZero needs no install and runs in any modern browser, including on a Mac or a phone, but solves postflop only and has no ICM. Both solve heads-up postflop spots, and both offer node locking, aggregated flop reports and a trainer.
It depends on what you want to study, and we make one of the three, so we won't rank them. For multiway pots or hand-history uploads, GTO Wizard has the feature; for preflop or ICM work, both GTO Wizard and PioSOLVER do, in different forms. For heads-up postflop spots on any device, free and without installing anything, GTOZero does the job — and the same spot can be run in more than one tool to compare.
Yes. Any modern browser on Windows, macOS or Linux runs it, with nothing to install. Custom solves use your CPU, so a faster machine solves faster. On iPhone and in Safari the browser allows only one thread, so custom solves are slower there.
Partly. Once the app is on your home screen, its stored Study Spots and trainer keep working without an internet connection. For custom solving, open the solver once while connected so the engine is downloaded; after that it can run offline.
Yes. The engine is WASM Postflop by Wataru Inariba (AGPL-3.0), whose original website notes that updates have ended. GTOZero continues the engine's development and publishes its full modified source under the same license.
The engine is based on WASM Postflop by Wataru Inariba, released under AGPL-3.0. GTOZero is a localized and extended version with hand review, precomputed flop data and the trainer, and publishes the full modified source under the same license.
Why it is free: we want more players to enjoy hold'em through objective, solver-based analysis rather than guesswork. We want solver study to be open to every player, so we made the solver free and added Study Spots, video guides and a trainer that teach you how to read the results and apply them to your own hands. The goal is simple — to help as many players as possible get better at hold'em.