Inaugural MOASEI Competition at AAMAS'2025: A Technical Report

TL;DR

The paper reports on the inaugural MOASEI competition at AAMAS 2025, evaluating multi-agent decision-making in open, dynamic environments using diverse AI techniques and providing insights into generalization and adaptation strategies.

Contribution

It introduces the MOASEI competition framework, new benchmarks for open-agent systems, and diverse solutions employing advanced AI architectures, advancing research in open-world multi-agent decision-making.

Findings

Promising strategies for generalization in open environments

Diverse AI solutions including graph neural networks and LLMs

Insights into robustness and responsiveness of agents

Abstract

We present the Methods for Open Agent Systems Evaluation Initiative (MOASEI) Competition, a multi-agent AI benchmarking event designed to evaluate decision-making under open-world conditions. Built on the free-range-zoo environment suite, MOASEI introduced dynamic, partially observable domains with agent and task openness--settings where entities may appear, disappear, or change behavior over time. The 2025 competition featured three tracks--Wildfire, Rideshare, and Cybersecurity--each highlighting distinct dimensions of openness and coordination complexity. Eleven teams from international institutions participated, with four of those teams submitting diverse solutions including graph neural networks, convolutional architectures, predictive modeling, and large language model--driven meta--optimization. Evaluation metrics centered on expected utility, robustness to perturbations, and responsiveness to environmental change. The results reveal promising strategies for generalization and adaptation in open environments, offering both empirical insight and infrastructure for future research. This report details the competition's design, findings, and contributions to the open-agent systems research community.