Overview

World Model Evaluation studies whether a model captures how an environment changes, especially when an agent acts or observations are missing. Approaches include open-loop prediction metrics, action-conditioned rollouts, closed-loop policy tests, physics-grounded diagnostics, and human or vision-language model judging. Benchmarks span robotic manipulation such as LIBERO and RoboTwin, driving such as NuPlan, and interactive GUI, web, and game environments. A central challenge is that visual realism and one-step accuracy often fail to predict physical consistency, long-term reliability, or planning success, so evaluation remains fragmented across tasks and horizons.

Latest papers

CardsList

245 papers40 papers in the last four weeks, down 25% on the four weeks before. 0.3% of all new papers.

  1. Plausible Videos Can Lead to Poor Decisions: Diagnosing and CURE-ing World Models for Planning

    Nov 17, 2026Raphael C. Kim, Michele Santacatterina, Jingsen Zhu +1World Model-Based PlanningAction-Conditioned World Models

  2. Every Camera Grows Up: Camera-Height Regression in Interactive Video World Models

    Nov 16, 2026Sheng Lu, Guangpin Tao, Tong Lu +1Interactive World ModelsWorld Model Evaluation

  3. The Flip Test: Video World Models Fail to Learn the World State

    Nov 15, 2026Mi Luo, Lijun Yu, Adams Wei Yu +10Video World ModelsWorld Model Learning

  4. Probing Where Surprise Happens: Localized Violation-of-Expectation Evaluation of Video World Models

    Nov 13, 2026Yanbin Xu, Ming bo CaiWorld Model EvaluationPhysical Reasoning in Video

  5. Recovered or Memorized? Investigating Dynamical Consistency and Faithfulness in World Model Latents

    Nov 13, 2026Jerome Francisco, Kyle Deng, George Flint +1World ModelsLatent World Models

  6. Beyond Prediction Gain: LASR as a Metric for Action-Grounded Latents in World Models

    Nov 13, 2026Yu Zhou, Changsong Lei, Yixiang Wei +6World Model EvaluationLatent Action Learning

  7. Does Your World Model Know Actions Have Consequences?

    Nov 11, 2026Ayush Gaggar, Jared Berry, Todd D. MurpheyWorld Model EvaluationAction-Conditioned World Models

  8. WanderBench: Can World Models Explore Dynamic Worlds from Video?

    Nov 9, 2026Yuhang Chen, Jiheng Li, Jiale Wang +2Video World ModelsWorld Model Evaluation

  9. How Many Exposures Does a World Model Need? Learning Unseen Game Mechanics at Test Time

    Nov 9, 2026Ku Onoda, Haruto Kawano, Shota Takashiro +1Game World ModelsWorld Model Learning

  10. An Evaluation of Tabular Foundation Models as Training-Free World Models

    Nov 9, 2026Xian Yeow Lee, Ryoichi Takase, Ahmed FarahatWorld Model LearningWorld Models for Robotics

  11. Multi-Objective Adaptation After Deployment: A Position on World Models for Remote Robot Teams

    Nov 7, 2026Jamie Santos, Kagan TumerRobot Policy AdaptationWorld Models for Robotics

  12. When Does Latent Prediction Error Support Physical Surprise Claims in Video World Models?

    Nov 1, 2026Anton Hristov, Aleksandar Slavov, Vasil VasilevVideo World ModelsWorld Model Evaluation

  13. Stand-In Effects and Pooling Amplification: Lessons from Biological World Models for Testing Action Effects

    Nov 1, 2026Bowen Xu, Boyu Chen, Jiaying LiWorld Model EvaluationStatistical Hypothesis Testing

  14. World Models Need Not Be Right to Rank Policies Right: Relative Bounds and Audit Limits

    Oct 31, 2026Ahanaf Hasan Ariq, Hasaan AhmadReinforcement LearningPolicy Evaluation

  15. Old Is Not False: Valid-Time Memory for Interactive Video World Models

    Oct 31, 2026Ziming Wang, Chuyue Gong, Hongliang Cao +4Memory-Augmented Video GenerationInteractive World Models

  16. Roz: A Taste Test for World Model Latents

    Oct 31, 2026Kenneth Ge, Dhruv Sheth, Kunvar ThamanWorld Model EvaluationWorld Models

  17. Spoofing the Latent State: Imperceptible Goal-Image Attacks Derail Planning in LeWorldModel

    Oct 30, 2026Kunal Bhosikar, Nicole Meng, Charu Sharma +1Adversarial Attacks on World ModelsLatent World Models

  18. One Verdict Is Not Enough: Regression Testing Learned World Models and Their Test Harnesses

    Oct 30, 2026Junyu Huo, Zihao Wan, Ziqi Mao +1DNN TestingWorld Model Evaluation