About

Free GTO Poker Range is a free study tool for 6-max No-Limit Hold'em preflop play. The strategies it shows come from a 7.5-billion-iteration CFR blueprint trained on the open-source DCFR-SOLVER (Rust, MIT). Average per-player exploitability against a reach-propagated vector best response is 0.307 % pot (≈ 6.1 bb/100). The full measurement methodology is on Methodology.

What you can do here

• Browse the preflop tree. 614 spots covering every open / 3-bet / 4-bet / all-in / fold branch at 7 bb open size. Click position tabs at the top to jump straight to any seat's RFI or defense range; click any cell to see the per-combo breakdown.
• Drill your decisions. The trainer deals you a random hand, plays out the surrounding action, and grades your move by GTO frequency and EV loss. Pause, replay, or change the starting spot (Open / vs Open / vs 3-bet / vs All-in) any time via Settings.
• See aggregate frequencies. For every spot the site shows the combo-weighted action mix (Fold / Call / Raise / All-in) and lets you toggle the grid between horizontal and vertical stripes.

What this site does NOT do

• Solve arbitrary spots on demand — we serve a fixed pre-trained blueprint.
• Cover postflop play yet — the current release is preflop only. Postflop EV is approximated via a 5 % empirical OOP positional-tax heuristic, calibrated on real flop boards (see Methodology).
• Support open sizes other than 7 bb yet — other sizings are on the roadmap.
• Provide real-money play assistance, RTA, or bot integration.
• Recommend or partner with online poker rooms.

Who built this

A solo project. The codebase, the training run, and this site are all maintained by one person learning poker theory deeply. For questions or feedback, see Contact.

Open about quality

We publish the algorithm (External-Sampling MCCFR with DCFR averaging α=1.5, β=0, γ=2.0), the game configuration (sizes, rake, abstraction), the empirical OOP-tax calibration, the full convergence history (avg per-player gap from 1.5 B to 7.5 B iterations), and the measured vector best-response gap on Methodology. The solver core is the open-source DCFR-SOLVER (MIT) by exinori, with our patches for vector BR and empirical tax.