GameHorizon Suite Provides Multi-Horizon Gameplay Evaluation
September 20, 2026
GameHorizon introduces a unified evaluation suite and an automated annotation pipeline for measuring AI gameplay across diverse temporal horizons. It includes GameHorizon-Data, a large-scale dataset for AAA games featuring language instructions and precise action control.
HOW THIS AFFECTS YOU
●
builderThe dataset provides a benchmark for testing agentic planning and instruction following in games.
●
researcherYou can now evaluate models on complex, long-horizon planning tasks in realistic game environments.