AgentOdyssey: Open-Ended Long-Horizon Text Game Generation for Test-Time Continual Learning Agents

For agents to learn continuously from interaction at test time, they must explore effectively, acquire new world knowledge and skills, retain relevant episodic experiences, and plan over long horizons. To evaluate these key abilities of test-time continual learning agents, we introduce AgentOdyssey, a novel evaluation framework that procedurally generates open-ended text games with rich entities, world dynamics, and long-horizon tasks. Critically, AgentOdyssey goes beyond the conventional assumption that learning does not occur at test time by placing agents in a continuous, long-horizon setting that interleaves learning and inference throughout deployment. We further propose a multifaceted evaluation methodology that measures game progress and offers diagnostic tests on world knowledge acquisition, episodic memory, object and action exploration, action diversity, and model cost. Evaluating diverse agent paradigms, our results reveal critical limits in agents' key abilities and factors that influence their meaningful horizon. Although performance scales with stronger base models, even the top agent remains far below human performance, leaving substantial headroom. Additionally, a focused study of agent test-time training finds that, although short-term memory is beneficial, agents trained with supervised fine-tuning still suffer from catastrophic forgetting, suggesting that effective parametric continual learning remains an open challenge for test-time agents.