Compiled Agency: Frontier General-Purpose Coding Agents Build Winning Game Players from Bare Interaction - from Flappy Bird to StarCraft II and Civilization
Frontier coding agents autonomously build winning game-playing programs from bare interaction, demonstrating a significant jump in general-purpose agentic capability.
Frontier coding agents autonomously build winning game-playing programs from bare interaction, demonstrating a significant jump in general-purpose agentic capability.
Auto-scored: A real capability escalation: general-purpose agents autonomously achieve strong performance across diverse games, signaling a jump in agentic competence.