google-deepmind/concordia

Health v2 · maintenance only

Adoption not part of the score

1647 stars · 359 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Concordia is a Python library for building generative agent-based social simulations, using a tabletop role-playing game pattern where a Game Master entity simulates environments and resolves natural-language actions from agent entities. It provides modular components, prefabs, and a simulation engine for constructing multi-agent models.

Use cases

• simulate social interactions between LLM-powered agents
• run generative agent-based models for social science research
• generate synthetic data for personalization
• evaluate real services through simulated usage
• study AI safety and ethics with simulated agents
• model economic or cognitive scenarios with multi-agent simulation

When to choose

• you need LLM-driven multi-agent simulations with natural-language actions
• you want a modular, component-based agent architecture with a Game Master pattern
• you are doing social science, economics, or AI safety research with simulated agents

When to avoid

• you need high-performance numeric agent-based modeling without LLMs
• you want a production chatbot or agent deployment framework rather than a research simulation library
• you have no access to an LLM API

Facets

library · maturity active

agent-framework simulation llm-inference data-generation artificial-intelligence simulation large-language-models data-science python cross-platform generative-agents agent-based-simulation social-simulation game-master multi-agent llm-driven-agents synthetic-data ai-agents

Member repositories

For agents

markdown · JSON · MCP: product_card(name="google-deepmind/concordia")