SpinGPT: A Large-Language-Model Approach to Playing Poker Correctly

2025-09-26

AI Summary

Traditional CFR-based algorithms face exponential computational complexity growth and fail to guarantee non-negative payoffs under Nash equilibrium in three-or-more-player poker, rendering them impractical for mainstream tournament formats such as Spin & Go. To address this, we propose SpinGPT—the first large language model (LLM) approach specifically designed for three-player imperfect-information poker. SpinGPT employs a novel two-stage training paradigm: (i) supervised fine-tuning on 320,000 high-quality human decision trajectories, followed by (ii) reinforcement learning optimization using 270,000 solver-generated expert demonstrations. Experimental results show that SpinGPT matches the solver’s action in 78% of decisions and achieves a statistically significant win rate of 13.4 ± 12.9 BB/100 (95% CI) against the strong baseline Slumbot. This work breaks long-standing theoretical and practical bottlenecks of classical algorithms in multi-player poker, establishing a new paradigm for LLM-based strategic reasoning in imperfect-information games.

Technology Category

• Game Theory and Economic Paradigms: Imperfect Information
• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Adversarial Agents

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

The Counterfactual Regret Minimization (CFR) algorithm and its variants have enabled the development of pokerbots capable of beating the best human players in heads-up (1v1) cash games and competing with them in six-player formats. However, CFR's computational complexity rises exponentially with the number of players. Furthermore, in games with three or more players, following Nash equilibrium no longer guarantees a non-losing outcome. These limitations, along with others, significantly restrict the applicability of CFR to the most popular formats: tournaments. Motivated by the recent success of Large Language Models (LLM) in chess and Diplomacy, we present SpinGPT, the first LLM tailored to Spin & Go, a popular three-player online poker format. SpinGPT is trained in two stages: (1) Supervised Fine-Tuning on 320k high-stakes expert decisions; (2) Reinforcement Learning on 270k solver-generated hands. Our results show that SpinGPT matches the solver's actions in 78% of decisions (tolerant accuracy). With a simple deep-stack heuristic, it achieves 13.4 +/- 12.9 BB/100 versus Slumbot in heads-up over 30,000 hands (95% CI). These results suggest that LLMs could be a new way to deal with multi-player imperfect-information games like poker.

Problem

• Overcoming CFR's exponential complexity in multi-player poker games
• Addressing Nash equilibrium limitations in three-player poker formats
• Developing LLM approach for imperfect-information multi-player games

Innovation

• LLM tailored to three-player poker format
• Two-stage training with supervised fine-tuning
• Reinforcement learning on solver-generated hands