A free poker study tool. It shows preflop ranges for tournaments and cash games, and solves postflop spots between two players right in your browser. No sign-up.
Most preflop ranges are labelled Estimated. They are not solved on the spot: they come from range charts and rules tuned so their overall frequencies match solver output at 20, 40 and 100 big blinds, with stacks in between blended from those.
At 15 big blinds or less, and whenever the stacks are short against the pot, ranges are labelled Nash: an all-in-or-fold game solved exactly for the stack, the ante and the rake.
With an ICM stage chosen, tournament ranges are priced in prize money instead of chips, for a 1,000-player field paying 150 with every stack equal.
Postflop is a real solve, not a lookup. A CFR solver compiled to WebAssembly runs on your device, starting from the two ranges that reached the flop, and stops when neither player can gain more than 0.5% of the pot by changing strategy on the tree it solved.
To finish in seconds, the flop solve looks ahead to a sample of turn and river cards with a smaller betting menu, and each street is solved again in full once its card is known. Fast flop looks at fewer cards: quicker and less accurate.
Postflop: measured against a solve of the complete tree with every candidate bet size, the default flop solve gives up about 1.7% of the pot on average, both players together (0.2% to 2.5% across seven test spots). Fast flop gives up about 3.3% (0.9% to 6.6%).
Preflop: hand by hand against public solver ranges, opening ranges agree on 87–94% of hands, ranges facing an open on 85–89%, and ranges facing a 3-bet on 73–75%. Spots the charts were tuned on are within about 3 percentage points of solver frequencies; others can be further off.
Read the output as a strong starting point for study, not as the last word of a full solver.
When a hand ends, an AI model can explain it in plain words from the solver's numbers for that hand. It is limited to 3 explanations a day per person, and the same line again is free. The explanation can misread the numbers or be wrong: the charts are the source.
There are no accounts. Your settings and language stay in your browser. Preflop requests carry only the game settings and the actions; postflop solving happens on your device. Visits are counted without cookies. When you ask for an AI explanation, that hand's solver numbers (the game settings, the actions played and the range statistics) are sent to Anthropic's Claude API to write it; nothing that identifies you goes with them. Its daily limit uses a random ID cookie and a scrambled form of your IP address.
gtoLite is an independent project for study. It is not affiliated with any poker site or solver company. Check your poker site's rules before using any tool while you play.
Found a range that looks wrong, or a bug? Open an issue on GitHub with the link to the spot.