AgentWorld is a deterministic spatial multi-agent simulation environment that includes an LLM-agent framework, verifiable reward mechanisms, and a connect-out benchmark platform. It is designed for reinforcement learning and agent research, allowing developers to test and evaluate autonomous agent behaviors in controlled, reproducible settings. The package is open-source under the MIT license.
In the AI & ML space, agentworld takes a focused approach. It focuses on creating reproducible, verifiable benchmarks for evaluating multi-agent LLM systems in spatial environments. agentworld is an open-source project aimed at AI researchers. The project is open source (MIT). agentworld is available on the command line and API.
agentworld first shipped in 2026. Among its 5 catalogued features are Multi-Agent Environment, LLM Agent Framework, and Verifiable Rewards.
Latest indexed changes and source events
Other apps tracked under the same category.
