OpenSpiel: A Framework for Reinforcement Learning in Games

2019/08/26 by Marc Lanctot, Lanctot, Marc, Edward Lockhart +50 · 19 citations
Computer Science · #Artificial Intelligence (cs.AI) #Artificial Intelligence in Games #Computer Science and Game Theory (cs.GT) #Evolutionary Algorithms and Applications #FOS: Computer and information sciences #Machine Learning (cs.LG) #Multiagent Systems (cs.MA) #Reinforcement Learning in Robotics

paper · pdf · doi:10.48550/arxiv.1908.09453

openalex publication_date 2019/08/26 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games. OpenSpiel supports n-player (single- and multi- agent) zero-sum, cooperative and general-sum, one-shot and sequential, strictly turn-taking and simultaneous-move, perfect and imperfect information games, as well as traditional multiagent environments such as (partially- and fully- observable) grid worlds and social dilemmas. OpenSpiel also includes tools to analyze learning dynamics and other common evaluation metrics. This document serves both as an overview of the code base and an introduction to the terminology, core concepts, and algorithms across the fields of reinforcement learning, computational game theory, and search.

Citations

• A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting
• The Complexity of Computing a Nash Equilibrium
• Artificial intelligence: a modern approach
• DeepStack: Expert-level artificial intelligence in heads-up no-limit poker
• The Hanabi challenge: A new frontier for AI research

Cited by

• Code-Space Response Oracles: Generating Interpretable Multi-Agent Policies with Large Language Models
• Observer, Not Player: Simulating Theory of Mind in LLMs through Game Observation
• Evaluating Generalization Capabilities of LLM-Based Agents in Mixed-Motive Scenarios Using Concordia
• Terra Nova: A Comprehensive Challenge Environment for Intelligent Agents
• Outbidding and Outbluffing Elite Humans: Mastering Liar's Poker via Self-Play and Reinforcement Learning
• Monopoly Deal: A Benchmark Environment for Bounded One-Sided Response Games
• Nash Policy Gradient: A Policy Gradient Method with Iteratively Refined Regularization for Finding Nash Equilibria
• Reward-Free Attacks in Multi-Agent Reinforcement Learning
• MARSHAL: Incentivizing Multi-Agent Reasoning via Self-Play with Strategic LLMs
• EARL: Efficient Agentic Reinforcement Learning Systems for Large Language Models
• Look-ahead Reasoning with a Learned Model in Imperfect Information Games
• GEM: A Gym for Agentic LLMs
• Beyond Game Theory Optimal: Profit-Maximizing Poker Agents for No-Limit Holdem
• SC2Tools: StarCraft II Toolset and Dataset API
• HLSMAC: A New StarCraft Multi-Agent Challenge for High-Level Strategic Decision-Making
• Using Reinforcement Learning to Optimize the Global and Local Crossing Number
• RLCard: A Toolkit for Reinforcement Learning in Card Games
• Swift for TensorFlow: A portable, flexible platform for deep learning
• Benchmarking Partial Observability in Reinforcement Learning with a Suite of Memory-Improvable Domains

Related

• Multi-agent Reinforcement Learning in OpenSpiel: A Reproduction Report
• OpenRL: A Unified Reinforcement Learning Framework
• On Reinforcement Learning for the Game of 2048
• Eden: A Unified Environment Framework for Booming Reinforcement Learning Algorithms
• Reinforcement Learning: From Algorithms To Foundation Models