ACT-Bench: Towards Action Controllable World Models for Autonomous Driving

World models have emerged as promising neural simulators for autonomous driving, with the potential to supplement scarce real-world data and enable closed-loop evaluations. However, current research primarily evaluates these models based on visual realism or downstream task performance, with limited focus on fidelity to specific action instructions. Although some studies address action fidelity, their evaluations rely on closed-source mechanisms, limiting reproducibility. To address this gap, we develop an open-access evaluation framework, ACT-Bench, for quantifying action fidelity, along with a baseline world model, Terra. Our framework includes a large-scale dataset pairing short context videos from nuScenes with corresponding future trajectories, which provide conditional inputs for generating future video frames and enable evaluation of action fidelity for executed motions. Leveraging this framework, we demonstrate that the state-of-the-art model does not fully adhere to given instructions, while Terra demonstrates better action fidelity. All components of our benchmark framework are publicly available to support future research.