GTOWizard

Can a language model compete against GTO Wizard AI? We put the world’s most popular LLMs to the test to see if they can actually make good decisions on the felt. Watch the full comparison 👇

The larger sizing is designed to fold out hands like AT-JTs, KJs, QJs, AQo, and A4/A5s, which still retain decent equity. A smaller 1/4 pot c-bet folds out practically nothing (only ~4% of hands), whereas a 3/4 pot bet increases folds to ~37%. SB overpairs benefit from the larger flop sizing, as it puts many of CO’s marginal hands and underpairs into an indifferent spot.

@Oxr00tz Hi, please dm us your email, thanks!

day #5 waiting for @GTOWizard to answer my email about getting an API key for benchmarking against their bot. benchmark.gtowizard.com

NL500 6-Max. 100bb deep. 3-bet pot, SB vs CO. Why does the SB prefer a 70% pot bet on 9♠6♥3♦? Drop your thoughts below 👇

Your postflop success is determined by how well you apply the mathematical foundations of preflop play. But do you understand where EV preflop comes from? TheWakko breaks down the mechanics behind preflop EV and how small adjustments before the flop can have a huge impact on your long-term results. Join us in this free coaching session at 6PM CET: app.gtowizard.com

@AllOrNubbin Hey! GTO Wizard is a web app - no download needed. Just head to gtowizard.com and log in from any browser. If the ‘Go to App’ button isn’t working, try clearing your cache or using a different browser. Let us know if you’re still having trouble.

Any advice on downloading @GTOWizard? The files that download when I download the Windows application is just a bunch of random zip files, none of which actually download or open the application itself. Even clicking the "Go to App" button on the site takes me nowhere.

Think open limping is a rookie mistake? Not in Mississippi Straddle games! In this article @thinkingpoker uses nodelocking in multiway preflop solutions to prove open limping can be the optimal strategy when the straddle is on. Learn why and when you should limp in straddled games & how to exploit those who don't. Read below! 👇

Try Multiway Preflop for free ➡️ gtow.pro/try_preflop_mu…

If you’re on the BTN facing a nitty player in the BB, how do you think the optimal RFI strategy changes? 🧐 We nodelocked this common spot for you. Swipe to learn the key adjustment and takeaway!

@TomDwan Great point, Tom. Running frontier LLMs at scale is expensive. That's why we use AIVAT, a variance reduction technique that achieves the same statistical significance in 10x fewer hands, so 5K is equivalent to ~50K raw hands.

This is cool. 5k obviously not enough hands though, you guys should know that. Can you run a new one with 50-100k hands 🙏🏻🙏🏻

@oinjusticeiro1 No. GTO Wizard AI plays a near-perfect Nash Equilibrium strategy. No human can beat that over a meaningful sample. The 4 bb/100 reference is what top pros win against other humans. Against GTO Wizard, they'd lose too.

@GTOWizard So a top PRO still beats GTO Wizard AI?

We benchmarked every major AI model at poker. GPT-5.4, Claude Opus 4.6, Gemini 3.1 Pro, Grok 4 and more. All played 5,000 hands of heads-up no-limit against our state-of-the-art poker agent. Every single one lost. Here's the full breakdown 🧵

We've been building the state-of-the-art in poker AI for years. Today we're opening it up. GTO Wizard Benchmark: ✅ Public API ✅ Live leaderboard ✅ AIVAT variance-reduced evaluation ✅ GitHub starter code + full documentation 🏆 Leaderboard: benchmark.gtowizard.com 📄 Paper: arxiv.org/abs/2603.23660 💻 GitHub: github.com/gtowizard-ai/r… Can your agent do better? We welcome all challengers. @OpenAI @ai_anthropic @GoogleDeepMind @xai

Why can't LLMs play poker? Poker is fundamentally different from math, coding, or trivia. Every hand involves: → Hidden information — you never see your opponent's cards → Balancing ranges across thousands of decision points → Long-term planning where each action shapes future streets → Opponent modeling under deep uncertainty These models can solve differential equations but can't construct a balanced river strategy. Even the best LLMs in the world still misread their own cards ~2% of the time — confusing suited for offsuit holdings.