apps-and-services
A comprehensive guide to GTO (Game Theory Optimal) poker solvers and AI trainers that use Counterfactual Regret Minimization (CFR) algorithms.
Table of Contents
- Cloud-Based Solvers
- Desktop Solvers
- Training Platforms (Play Against AI)
- Mobile Apps
- Free/Open Source Options
- Comparison Summary
Cloud-Based Solvers
GTO Wizard
Website: gtowizard.com
The leading cloud-based GTO solver with AI-powered instant solving.
| Feature | Details |
|---|---|
| Solving Speed | ~3 seconds per street for up to 200bb |
| Accuracy | 0.1-0.3% Nash Distance |
| Algorithm | Neural Network + CFR hybrid |
| Training Mode | Yes, play against GTO AI |
| Pricing | Subscription-based |
Key Features:
- Massive pre-solved library + custom solving
- GTO Trainer for practice
- Upgraded to QRE (Quantal Response Equilibrium) in April 2025
- Beat Slumbot for 19.4bb/100 over 150k hands
Deepsolver
Website: deepsolver.com
Neural network-based solver with on-demand computation.
| Feature | Details |
|---|---|
| Solving Speed | Seconds (“blinding speed”) |
| Training Mode | Yes, up to 4 tables simultaneously |
| Algorithm | Neural network solver |
| Pricing | Subscription-based |
Key Features:
- Computes trees on-demand (not pre-solved library)
- Fully customizable GTO Trainer
- Sizing optimization tools
Desktop Solvers
PioSolver
Website: piosolver.com
The gold standard desktop solver, widely used by professionals.
| Feature | Details |
|---|---|
| Type | Local software (Windows) |
| Algorithm | CFR |
| Pricing | Starting from $249 |
| Hardware | Requires powerful CPU + RAM |
Key Features:
- Deepest analysis capabilities
- Custom ranges, bet sizes, rake structures
- Steepest learning curve
- Industry standard for serious study
GTO+
Desktop solver alternative to PioSolver with lower price point.
MonkerSolver
Specialized for multi-way pots and PLO.
Training Platforms (Play Against AI)
These platforms let you play hands against a GTO opponent with real-time feedback.
GTO Wizard Trainer
Part of GTO Wizard subscription. Play fictional hands against GTO AI.
WPT GTO Trainer
Website: gto.learnwpt.com
| Feature | Details |
|---|---|
| Solved Spots | 4 billion+ |
| Feedback | Real-time EV loss analysis |
| Focus | Tournament and cash games |
GTOBase Trainer
Website: gtobase.com
| Feature | Details |
|---|---|
| Training Modes | Multiple modes for different skills |
| Algorithm | SimplePoker solving algorithms |
| Feedback | Real-time mistake analysis |
PokerSnowie
Good for beginners. Assumes all opponents play GTO.
Poker Scientist
Website: poker-scientist.com
Practice against GTO-Bot with tight or aggressive logic options.
Poker Genius
Website: poker-genius.com
Play against multiple AI bots (Jagbot, Pokibot, Simbot, etc.). Supports:
- All Hold’em formats (Full Ring, 6-max, HU)
- Various tournament structures
- Custom bot development
Mobile Apps
Poker Solver+ (iOS)
App Store: Poker Solver+
GTO postflop lookup in your pocket. No complex tree building required.
GTO Ranges+ (Android)
Play Store: GTO Ranges+
Professionally solved multi-way ranges for cash games, MTTs, and Spin & Gos.
Free/Open Source Options
PeakGTO
Website: pokercoaching.com/peakgto
Completely free solver by Pokercoaching.com with:
- Full solver functionality
- Training mode with feedback
- Comparable to paid alternatives
WASM Postflop
GitHub: b-inary/wasm-postflop
Free, open-source browser-based solver.
| Feature | Details |
|---|---|
| Algorithm | Discounted CFR |
| Platform | Web browser |
| Status | Development suspended (Oct 2023) |
| Features | Multithreading support |
Pluribus Implementations
GitHub: pluribus-poker-AI
Open source implementation of key Pluribus concepts.
| Feature | Details |
|---|---|
| Language | Python 3.7+ |
| Algorithm | Monte Carlo CFR |
| License | GPL |
| Status | Research/prototype quality |
Multiple forks available. Implements blueprint strategy creation using MCCFR.
Other Open Source
| Project | Description |
|---|---|
| Gongsta/Poker-AI | Resources and implementation collection |
| tansey/pycfr | Python CFR implementation |
| doas3140/PyStack | DeepStack Python implementation |
Comparison Summary
| Tool | Type | Play vs AI | Price | Best For |
|---|---|---|---|---|
| GTO Wizard | Cloud | Yes | $$$ | Serious study + training |
| Deepsolver | Cloud | Yes | $$$ | Fast custom solves |
| PioSolver | Desktop | No | $249+ | Deep analysis |
| WPT GTO Trainer | Cloud | Yes | $$ | Tournament focus |
| GTOBase | Cloud | Yes | $$ | Balanced features |
| PeakGTO | Cloud | Yes | Free | Budget learners |
| WASM Postflop | Browser | No | Free | Quick analysis |
| Poker Genius | Desktop | Yes | $ | Bot variety |
Technical Notes
How CFR Works
CFR explores every decision point for every hand combination each iteration. With each iteration, strategy converges toward Nash Equilibrium. Modern solvers combine CFR with:
- Neural Networks - Accelerate solving (GTO Wizard AI, Deepsolver)
- Discounted Regret - Faster convergence (Pluribus technique)
- Abstractions - Handle game tree size (card bucketing, bet sizing)
Algorithm Variants
| Variant | Description |
|---|---|
| Vanilla CFR | Original 2007 algorithm |
| CFR+ | Faster convergence, used in desktop solvers |
| MCCFR | Monte Carlo sampling for large games |
| Discounted CFR | Weights recent iterations higher |
| Linear CFR | Linear weighting scheme |