The four kinds of tool, and what each is for
Everything sold as a "GTO tool" falls into one of four categories. They solve different problems and are only loosely substitutes for one another.
1. Precomputed chart libraries
A stored set of solved preflop ranges you look up. No computation happens when you use one — the answer was calculated in advance. This is the right tool for every preflop question, and preflop is the only street you play in every hand. The chart shelf here is one of these and is free.
2. Postflop solvers
These actually run counterfactual regret minimisation on a spot you define: ranges, board, pot, stack, bet sizes. The right tool when you want to know what happens on a specific texture, and the only one that will answer a question nobody precomputed.
3. Trainers
Deal you a spot, take your decision, and grade it against a solve. A different product from a solver even when a solver is inside it — a solver answers questions, a trainer builds recall. Most players who buy a solver actually wanted this.
4. Hand-history analysers
Take hands you have already played and score them in bulk. Useful once you have a real sample, useless before that, and dependent on playing somewhere that gives you hand histories.
Which problem do you actually have?
Match the symptom to the category before looking at any product:
| What you notice | What you need | What it costs |
|---|---|---|
| "I don’t know what to open from the cutoff." | A chart library. | Nothing. This is solved and freely available. |
| "I know the chart but freeze at the table." | A trainer, under time pressure. | Time, mostly. |
| "I keep losing big pots on paired boards." | A postflop solver, on that texture. | A free browser solver handles this. |
| "I don’t know which of my spots is worst." | A hand-history analyser or a leak report. | Needs a sample first. |
| "I want to design a bet-size menu for deep 3-bet pots." | A serious desktop solver. | This is the case that justifies paying. |
Reading down that table honestly is uncomfortable, because the second row is where most players are and it is the one no purchase fixes.
What free solvers genuinely cannot do
The limitation of a free or browser-based solver is almost never accuracy. CFR+ converges to the same equilibrium regardless of what it runs on; a converged solve is a converged solve. What you are buying with a paid desktop tool is tree size, and the specific things a large tree buys are:
- Many bet sizes per street. Free tools give you two or three. This matters if you are designing a sizing strategy, and does not if you want to know whether to bet.
- Full preflop trees at depth. Solving preflop from scratch with realistic sizing is a large-machine job. It is also unnecessary, because the answers are stable and already published as charts.
- Bulk automation. Solving hundreds of boards unattended and aggregating the results.
- Node locking. Fixing one player’s strategy to something wrong on purpose, then solving the best response — the formal version of exploitative play. This is the most genuinely valuable paid feature and the least used.
If none of those four appear in your study routine, a free browser solver is not a compromise. It is the same mathematics with a smaller menu.
Accuracy is not the paywall. Tree size and automation are. A converged three-size solve of the flop you actually misplayed is worth more than an unrun twelve-size solve of one you did not.
Browser or desktop
WebAssembly closed most of the arithmetic gap, so this is now a question about workflow rather than capability.
- Browser — nothing to install, works on a machine you do not administer, and if the solve runs client-side the hands never leave your computer. Best for looking something up in the moment. The solver here runs in a Web Worker in your own browser with no account and no upload.
- Desktop — larger trees, bulk runs, saved libraries you build up over months. Best if solving is a scheduled activity rather than an interruption.
One point worth checking about any browser tool: whether the solve runs on your machine or on their server. Server-side is not automatically bad, but it means your hands and your study are being uploaded, and it is worth knowing which one you have chosen.
The bottleneck no solver removes
The reason so many solver subscriptions go unused after a month is that the tool answers a question you asked once, and the problem it was bought for is not a question. It is a recall failure under time pressure.
You can know the correct continuation-betting frequency on a dry ace-high flop, having read it off a solve, and still get it wrong at a table, because a table gives you a few seconds, no grid, and a hand you feel something about. Closing that gap requires being dealt the spot repeatedly and being corrected immediately — which is a trainer’s job, not a solver’s.
- Use a chart library to learn what the ranges are. Free, and finished in a week or two.
- Use a trainer to make them survive a clock. This is where the time goes, and it is the step most people skip.
- Use a solver for the specific spots that keep costing you money after that. Occasional, targeted, and rarely worth a subscription until it becomes frequent.
That is the order the seven-chapter bootcamp runs in, with the solve happening on your own machine and every decision priced in big blinds. Chapter one needs no account or card, chapter two needs a free account, and the leak report tells you which spots are the ones worth opening a solver for.
Frequently asked questions
Is there a genuinely free poker solver?
Yes. GTO Academy runs a CFR+ postflop solver entirely in your browser with no account and no download, and publishes solved preflop charts free for every game type, position and stack depth. Free tools are limited by tree size — how many bet sizes and how deep a tree they can handle — rather than by the accuracy of the solve.
Are paid poker solvers more accurate than free ones?
No. CFR+ converges on the same Nash equilibrium regardless of what it runs on, so a converged solve is a converged solve. What paid desktop tools buy is tree size, bulk automation, full preflop solving and node locking — capability, not correctness.
Do I need a solver to improve at poker?
Usually not at first. Preflop is already solved and published as free charts, and the most common problem — knowing the right play but not recalling it under time pressure — is fixed by a trainer rather than a solver. A solver becomes worth it once you have identified specific postflop spots that keep costing you money.
What is node locking and do I need it?
Node locking fixes one player’s strategy to something deliberately wrong, then solves the best response to it — the formal version of exploitative play. It is the most genuinely valuable feature behind a paywall and one of the least used. You need it if you are building strategies against a specific opponent pool, and not otherwise.
Is a browser solver safe to use?
It depends on where the solve runs. A solver that computes in your own browser, as GTO Academy’s does in a Web Worker, never uploads the hands, ranges or boards you enter. A server-side tool sends them to be computed elsewhere, which is worth knowing about before you decide.