pith. sign in

arxiv: 2605.25874 · v1 · pith:MN2WZK4Cnew · submitted 2026-05-25 · 💻 cs.CV

WBench: A Comprehensive Multi-turn Benchmark for Interactive Video World Model Evaluation

Kaining Ying , Hengrui Hu , Siyu Ren , Jiamu Li , Fengjiao Chen , Ziwen Wang , Xuezhi Cao , Xunliang Cai
show 1 more author
Henghui Ding
This is my paper

Pith reviewed 2026-06-29 22:54 UTC · model grok-4.3

classification 💻 cs.CV
keywords interactive world modelsvideo generation benchmarksmulti-turn evaluationphysics compliancevideo quality assessmentworld model diagnostics
0
0 comments X

The pith

WBench introduces a multi-turn benchmark evaluating interactive video world models on five dimensions and finds no model performs strongly across all.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents WBench to create a unified standard for testing interactive world models that current benchmarks cover only partially. It defines evaluation along video quality, setting adherence, interaction adherence, consistency, and physics compliance using 289 test cases with 1058 turns that include varied scenes, styles, perspectives, and four interaction types. Metrics combine specialist vision models with large multimodal models into 22 automatic sub-metrics, all checked against human judgments. When applied to 20 state-of-the-art models the results show each model has distinct strengths and weaknesses with none succeeding uniformly. The work supplies per-model diagnostics on open challenges in these interactive settings.

Core claim

WBench supplies 289 test cases and 1058 interaction turns that specify world settings and multi-turn sequences covering navigation, subject action, event editing, and perspective switching, with unified support for text, 6-DoF pose, and discrete-action controls; its 22 automatic sub-metrics, validated against human judgments, demonstrate that no model among the 20 evaluated performs strongly across the five dimensions of video quality, setting adherence, interaction adherence, consistency, and physics compliance.

What carries the argument

The WBench benchmark itself, structured around its five evaluation dimensions and the set of 22 automatic sub-metrics that combine specialist vision models with large multimodal models.

If this is right

  • Systematic multi-turn testing across navigation, action, editing, and perspective changes is required to expose consistency and adherence failures invisible in single-turn evaluations.
  • Unified support for text, pose, and discrete-action inputs allows direct comparison of models with different native control interfaces.
  • Diagnostic per-dimension scores reveal characteristic model weaknesses that can guide targeted improvements rather than uniform scaling.
  • Validated automatic metrics enable scalable evaluation without relying solely on costly human annotation for every test case.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The benchmark's coverage of first- and third-person perspectives may help identify viewpoint-dependent failure modes that transfer to embodied agents.
  • If the five dimensions prove partially independent, future architectures may need explicit modular components rather than end-to-end training alone.
  • Extending the test cases to longer interaction sequences could surface compounding physics violations not captured in the current 1058 turns.
  • The finding that no model dominates all dimensions suggests that leaderboard rankings based on single metrics will continue to mislead development priorities.

Load-bearing premise

The 22 automatic sub-metrics accurately reflect the five dimensions when validated against human judgments.

What would settle it

A model that scores at the top across all five dimensions on the full set of WBench test cases, or a large mismatch between the automatic sub-metric rankings and fresh human judgments on the same outputs.

Figures

Figures reproduced from arXiv: 2605.25874 by Fengjiao Chen, Henghui Ding, Hengrui Hu, Jiamu Li, Kaining Ying, Siyu Ren, Xuezhi Cao, Xunliang Cai, Ziwen Wang.

Figure 1
Figure 1. Figure 1: Overview of WBENCH. Top: a multi-turn case with navigation, subject action, event editing, and perspective switching. Bottom: the benchmark design, including world settings, interaction taxonomy, unified navigation control, and evaluation over video quality, setting adherence, interaction adherence (navigation and semantic interactions), consistency, and physics compliance. ABSTRACT Interactive world model… view at source ↗
Figure 2
Figure 2. Figure 2: Dataset composition of WBENCH across eight axes. We discuss these in Section 3.2. an initial frame form the input to each evaluated model. Initial frames are generated by Nano Banana 2 [63] and GPT-Image-1.5 [64], supplemented by web-collected and manually captured images. All initial frames undergo manual verification for quality control. Interactions. Each case specifies a multi-turn interaction sequence… view at source ↗
Figure 3
Figure 3. Figure 3: Cross-dimension correlation and per-setting deviation analysis. [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Per-turn performance degradation. T4+ aggregates all turns from the 4 [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Spearman ρ between per-model human win rates (x-axis) and automated WBENCH scores (y-axis) across ten evaluation aspects. All aspects achieve ρ ≥ 0.94, with four reaching ρ = 1.00. compound across steps. Dedicated world models are more robust: HY-World 1.5 degrades much less than Kling 3.0, suggesting that explicit geometric control better preserves spatial state than text-based prompting. Event editing an… view at source ↗
Figure 6
Figure 6. Figure 6: Thumbnail gallery of all cases in WBENCH. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Scene and style coverage. Two categories per row are presented, each shown as a photorealistic/stylized pair [PITH_FULL_IMAGE:figures/full_fig_p021_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Style gallery. Representative initial frames spanning the rendering styles covered by [PITH_FULL_IMAGE:figures/full_fig_p022_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Perspective gallery. Cases are grouped into three rows by perspective type: disembodied first-person with [PITH_FULL_IMAGE:figures/full_fig_p022_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Subject gallery. One column per subject category, covering human, animal, vehicle, robot, and other objects. [PITH_FULL_IMAGE:figures/full_fig_p023_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Perspective-switching taxonomy showcase. One representative case per sub-type, covering same-subject [PITH_FULL_IMAGE:figures/full_fig_p024_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Navigation action definition. Illustration of how each WASD and arrow key maps to a physical motion [PITH_FULL_IMAGE:figures/full_fig_p026_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Distribution of navigation test cases across direction, scene type, and control interface. [PITH_FULL_IMAGE:figures/full_fig_p026_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Automated web-based evaluation pipeline for Genie 3 and Happy Oyster. Given an initial image and a text [PITH_FULL_IMAGE:figures/full_fig_p030_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: A four-turn navigation case (T1: W, T2: →, T3: →, T4: ←) across three models. Frames 1–3 correspond to T1, 4–6 to T2, 7–9 to T3, and 10–12 to T4. Happy Oyster and HY-World 1.5 follow the instructed directions correctly, while HY-Video 1.5 reverses the rotation in T2 and T3. 35 [PITH_FULL_IMAGE:figures/full_fig_p035_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Adaptive GT construction for the same case. The GT trajectory adapts to each model’s predicted motion [PITH_FULL_IMAGE:figures/full_fig_p036_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Qualitative comparisons on event editing. [PITH_FULL_IMAGE:figures/full_fig_p037_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Qualitative comparisons on subject action. [PITH_FULL_IMAGE:figures/full_fig_p038_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Qualitative comparisons on perspective switching. [PITH_FULL_IMAGE:figures/full_fig_p039_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: Qualitative comparisons on spatial consistency. The gated score penalises the static scene. [PITH_FULL_IMAGE:figures/full_fig_p040_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Qualitative comparisons on physics compliance. [PITH_FULL_IMAGE:figures/full_fig_p045_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: Causal Fidelity decomposed along the seven Track 2 sub-dimensions. [PITH_FULL_IMAGE:figures/full_fig_p046_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: Human-preference annotation platform. Each task is a blind pairwise comparison between two models, with [PITH_FULL_IMAGE:figures/full_fig_p050_23.png] view at source ↗
read the original abstract

Interactive world models are advancing rapidly, yet existing benchmarks cover only part of the required competencies, leaving no unified standard for systematic evaluation. To fill this gap, we introduce WBench, a comprehensive multi-turn benchmark for interactive world model evaluation along five dimensions, namely video quality, setting adherence, interaction adherence, consistency, and physics compliance. WBench contains 289 test cases and 1,058 interaction turns, where each case specifies a world setting and a multi-turn interaction sequence, covering diverse scenes, styles, subjects, and both first- and third-person perspectives, together with four interaction types, including navigation, subject action, event editing, and perspective switching. For navigation, WBench unifies text, 6-DoF pose, and discrete-action control, enabling evaluation of models with different native input interfaces. Evaluation uses 22 automatic sub-metrics that combine specialist vision models with large multimodal models, and all metrics are validated against human judgments. Across 20 state-of-the-art models, we find that no single model performs strongly across all dimensions. We provide detailed diagnostic insights into the characteristic strengths, weaknesses, and open challenges of each model. Code and data are available at https://github.com/meituan-longcat/WBench.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript introduces WBench, a multi-turn benchmark for interactive video world model evaluation along five dimensions (video quality, setting adherence, interaction adherence, consistency, and physics compliance). It contains 289 test cases and 1,058 interaction turns covering diverse scenes, styles, subjects, perspectives, and four interaction types, with unified control interfaces for navigation. Evaluation relies on 22 automatic sub-metrics that combine specialist vision models with large multimodal models; these metrics are stated to be validated against human judgments. Across 20 state-of-the-art models, the paper reports that no single model performs strongly across all dimensions and provides diagnostic insights into model strengths and weaknesses. Code and data are released at a public GitHub repository.

Significance. If the automatic metrics are shown to be faithful proxies, WBench would fill an important gap by providing a unified, multi-turn standard for interactive world models and would usefully document complementary model capabilities. The public release of code and data is a clear strength that supports reproducibility.

major comments (1)
  1. [Abstract] Abstract: the statement that 'all metrics are validated against human judgments' supplies no quantitative support (correlation coefficients, number of raters, inter-rater agreement, or per-dimension breakdown). Because the headline result that 'no single model performs strongly across all dimensions' is produced entirely by the 22 automatic sub-metrics, the missing validation statistics are load-bearing for the central claim.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the major comment point by point below and commit to revisions that directly strengthen the presentation of our validation claims.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the statement that 'all metrics are validated against human judgments' supplies no quantitative support (correlation coefficients, number of raters, inter-rater agreement, or per-dimension breakdown). Because the headline result that 'no single model performs strongly across all dimensions' is produced entirely by the 22 automatic sub-metrics, the missing validation statistics are load-bearing for the central claim.

    Authors: We agree that the abstract should supply quantitative support for the claim that the 22 automatic sub-metrics are validated against human judgments. The manuscript body contains the relevant validation details (including correlations, rater counts, agreement metrics, and per-dimension results), but these are not summarized in the abstract. We will revise the abstract to include a concise quantitative statement of the validation results, thereby making the load-bearing evidence for the headline finding explicit and transparent. revision: yes

Circularity Check

0 steps flagged

No circularity: benchmark construction is external to evaluated models

full rationale

The paper introduces WBench as an independent benchmark with 289 test cases and 22 automatic sub-metrics for five evaluation dimensions. The central result (no model strong across all dimensions) is produced by applying these metrics to 20 external models. No self-definitional equations, fitted parameters renamed as predictions, or load-bearing self-citations appear in the provided text; the metrics are presented as combining specialist vision models with large multimodal models and validated externally against human judgments. The work is therefore self-contained as an empirical evaluation framework rather than a closed derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Benchmark introduction paper; no mathematical derivations, free parameters, or invented entities involved.

pith-pipeline@v0.9.1-grok · 5780 in / 992 out tokens · 32700 ms · 2026-06-29T22:54:57.172034+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Current World Models Lack a Persistent State Core

    cs.CV 2026-06 unverdicted novelty 6.0

    Current world models fail to evolve internal state when unobserved and instead resume scenes at the last observed state, as diagnosed by the new WRBench benchmark across 23 models and 9600 videos.

  2. Einstein World Models

    cs.AI 2026-06 unverdicted novelty 5.0

    Einstein World Models integrate visual rollouts from a callable world-module into LLM reasoning traces to support complex thought beyond language.

  3. WorldOlympiad: Can Your World Model Survive a Triathlon?

    cs.CV 2026-06 unverdicted novelty 5.0

    WorldOlympiad is a new benchmark decomposing world-model evaluation into physical, geometry, and interaction tracks using segmentation, MLLM judges, Gaussian splatting, and action prompts on diverse scenarios.

Reference graph

Works this paper leans on

83 extracted references · 50 canonical work pages · cited by 3 Pith papers · 27 internal anchors

  1. [1]

    Denoising diffusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020

    2020

  2. [2]

    HunyuanVideo: A Systematic Framework For Large Video Generative Models

    Weijie Kong, Qi Tian, Zijian Zhang, Rox Min, Zuozhuo Dai, Jin Zhou, Jiangfeng Xiong, Xin Li, Bo Wu, Jianwei Zhang, et al. Hunyuanvideo: A systematic framework for large video generative models.arXiv preprint arXiv:2412.03603, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  3. [4]

    Movie Gen: A Cast of Media Foundation Models

    Adam Polyak, Amit Zohar, Andrew Brown, Andros Tjandra, Animesh Sinha, Ann Lee, Apoorv Vyas, Bowen Shi, Chih-Yao Ma, Ching-Yao Chuang, et al. Movie gen: A cast of media foundation models.arXiv preprint arXiv:2410.13720, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  4. [5]

    Video generation models as world simulators.OpenAI Blog, 1(8):1, 2024

    Tim Brooks, Bill Peebles, Connor Holmes, Will DePue, Yufei Guo, Leo Jing, David Schnurr, Joe Taylor, Troy Luhman, Eric Luhman, et al. Video generation models as world simulators.OpenAI Blog, 1(8):1, 2024

    2024

  5. [6]

    Genie: Generative interactive environments

    Jake Bruce, Michael D Dennis, Ashley Edwards, Jack Parker-Holder, Yuge Shi, Edward Hughes, Matthew Lai, Aditi Mavalankar, Richie Steigerwald, Chris Apps, et al. Genie: Generative interactive environments. InForty-first International Conference on Machine Learning, 2024

    2024

  6. [7]

    Genie 2: A large-scale foundation world model

    Jack Parker-Holder, Philip Ball, Jake Bruce, Vibhavari Dasagi, Kristian Holsheimer, Christos Kaplanis, Alexandre Moufarek, Guy Scully, Jeremy Shar, Jimmy Shi, et al. Genie 2: A large-scale foundation world model. https://deepmind.google/ discover/blog/genie-2-a-large-scale-foundation-world-model/, 2024. 11 WBENCH: A Benchmark for Interactive Video World M...

    2024

  7. [8]

    Philip J. Ball, Jakob Bauer, Frank Belletti, Bethanie Brownfield, Ariel Ephrat, Shlomi Fruchter, Agrim Gupta, Kristian Holsheimer, Aleksander Holynski, Jiri Hron, Christos Kaplanis, Marjorie Limont, Matt McGill, Yanko Oliveira, Jack Parker- Holder, Frank Perbet, Guy Scully, Jeremy Shar, Stephen Spencer, Omer Tov, Ruben Villegas, Emma Wang, Jessica Yung, C...

    2025

  8. [9]

    Matrix-game 2.0: An open-source real-time and streaming interactive world model

    Xianglong He, Chunli Peng, Zexiang Liu, Boyang Wang, Yifan Zhang, Qi Cui, Fei Kang, Biao Jiang, Mengyin An, Yangyang Ren, et al. Matrix-game 2.0: An open-source real-time and streaming interactive world model.arXiv preprint arXiv:2508.13009, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  9. [10]

    Matrix-Game 3.0: Real-Time and Streaming Interactive World Model with Long-Horizon Memory

    Zile Wang, Zexiang Liu, Jaixing Li, Kaichen Huang, Baixin Xu, Fei Kang, Mengyin An, Peiyu Wang, Biao Jiang, Yichen Wei, et al. Matrix-game 3.0: Real-time and streaming interactive world model with long-horizon memory.arXiv preprint arXiv:2604.08995, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  10. [11]

    Hunyuan- gamecraft: High-dynamic interactive game video generation with hybrid history condition.arXiv preprint arXiv:2506.17201, 2 (3):6, 2025

    Jiaqi Li, Junshu Tang, Zhiyong Xu, Longhuang Wu, Yuan Zhou, Shuai Shao, Tianbao Yu, Zhiguo Cao, and Qinglin Lu. Hunyuan- gamecraft: High-dynamic interactive game video generation with hybrid history condition.arXiv preprint arXiv:2506.17201, 2 (3):6, 2025

    work page arXiv 2025

  11. [12]

    Hunyuan-gamecraft-2: Instruction-following interactive game world model.arXiv preprint arXiv:2511.23429, 2025

    Junshu Tang, Jiacheng Liu, Jiaqi Li, Longhuang Wu, Haoyu Yang, Penghao Zhao, Siruis Gong, Xiang Yuan, Shuai Shao, Linfeng Zhang, et al. Hunyuan-gamecraft-2: Instruction-following interactive game world model.arXiv preprint arXiv:2511.23429, 2025

    work page arXiv 2025

  12. [13]

    GAIA-1: A Generative World Model for Autonomous Driving

    Anthony Hu, Lloyd Russell, Hudson Yeo, Zak Murez, George Fedoseev, Alex Kendall, Jamie Shotton, and Gianluca Corrado. Gaia-1: A generative world model for autonomous driving.arXiv preprint arXiv:2309.17080, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  13. [14]

    Vista: A generalizable driving world model with high fidelity and versatile controllability.Advances in Neural Information Processing Systems, 37:91560–91596, 2024

    Shenyuan Gao, Jiazhi Yang, Li Chen, Kashyap Chitta, Yihang Qiu, Andreas Geiger, Jun Zhang, and Hongyang Li. Vista: A generalizable driving world model with high fidelity and versatile controllability.Advances in Neural Information Processing Systems, 37:91560–91596, 2024

    2024

  14. [15]

    Learning Interactive Real-World Simulators

    Sherry Yang, Yilun Du, Kamyar Ghasemipour, Jonathan Tompson, Leslie Kaelbling, Dale Schuurmans, and Pieter Abbeel. Learning interactive real-world simulators.arXiv preprint arXiv:2310.06114, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  15. [16]

    Irasim: Learning interactive real-robot action simulators.arXiv preprint arXiv:2406.14540, 1(2):3, 2024

    Fangqi Zhu, Hongtao Wu, Song Guo, Yuxiao Liu, Chilam Cheang, and Tao Kong. Irasim: Learning interactive real-robot action simulators.arXiv preprint arXiv:2406.14540, 1(2):3, 2024

    work page arXiv 2024

  16. [17]

    Hy-world 1.5: A systematic framework for interactive world modeling with real-time latency and geometric consistency.arXiv preprint, 2025

    Team HunyuanWorld. Hy-world 1.5: A systematic framework for interactive world modeling with real-time latency and geometric consistency.arXiv preprint, 2025

    2025

  17. [18]

    Yume: An interactive world generation model.arXiv preprint arXiv:2507.17744, 2025

    Xiaofeng Mao, Shaoheng Lin, Zhen Li, Chuanhao Li, Wenshuo Peng, Tong He, Jiangmiao Pang, Mingmin Chi, Yu Qiao, and Kaipeng Zhang. Yume: An interactive world generation model.arXiv preprint arXiv:2507.17744, 2025

    work page arXiv 2025

  18. [19]

    Yume-1.5: A text-controlled interactive world generation model.arXiv preprint arXiv:2512.22096, 2025

    Xiaofeng Mao, Zhen Li, Chuanhao Li, Xiaojie Xu, Kaining Ying, Tong He, Jiangmiao Pang, Yu Qiao, and Kaipeng Zhang. Yume-1.5: A text-controlled interactive world generation model.arXiv preprint arXiv:2512.22096, 2025

    work page arXiv 2025

  19. [20]

    Advancing Open-source World Models

    Robbyant Team, Zelin Gao, Qiuyu Wang, Yanhong Zeng, Jiapeng Zhu, Ka Leong Cheng, Yixuan Li, Hanlin Wang, Yinghao Xu, Shuailei Ma, et al. Advancing open-source world models.arXiv preprint arXiv:2601.20540, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  20. [21]

    Vbench: Comprehensive benchmark suite for video generative models

    Ziqi Huang, Yinan He, Jiashuo Yu, Fan Zhang, Chenyang Si, Yuming Jiang, Yuanhan Zhang, Tianxing Wu, Qingyang Jin, Nattapol Chanpaisit, et al. Vbench: Comprehensive benchmark suite for video generative models. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 21807–21818, 2024

    2024

  21. [22]

    VBench-2.0: Advancing Video Generation Benchmark Suite for Intrinsic Faithfulness

    Dian Zheng, Ziqi Huang, Hongbo Liu, Kai Zou, Yinan He, Fan Zhang, Lulu Gu, Yuanhan Zhang, Jingwen He, Wei-Shi Zheng, et al. Vbench-2.0: Advancing video generation benchmark suite for intrinsic faithfulness.arXiv preprint arXiv:2503.21755, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  22. [23]

    WorldMark: A Unified Benchmark Suite for Interactive Video World Models

    Xiaojie Xu, Zhengyuan Lin, Kang He, Yukang Feng, Xiaofeng Mao, Yuanyang Yin, Kaipeng Zhang, and Yongtao Ge. Worldmark: A unified benchmark suite for interactive video world models.arXiv preprint arXiv:2604.21686, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  23. [24]

    Mind: Benchmarking memory consistency and action control in world models.arXiv preprint arXiv:2602.08025, 2026

    Yixuan Ye, Xuanyu Lu, Yuxin Jiang, Yuchao Gu, Rui Zhao, Qiwei Liang, Jiachun Pan, Fengda Zhang, Weijia Wu, and Alex Jinpeng Wang. Mind: Benchmarking memory consistency and action control in world models.arXiv preprint arXiv:2602.08025, 2026

    work page arXiv 2026

  24. [25]

    Omni-worldbench: Towards a comprehensive interaction-centric evaluation for world models.arXiv preprint arXiv:2603.22212, 2026

    Meiqi Wu, Zhixin Cai, Fufangchen Zhao, Xiaokun Feng, Rujing Dang, Bingze Song, Ruitian Tian, Jiashu Zhu, Jiachen Lei, Hao Dou, et al. Omni-worldbench: Towards a comprehensive interaction-centric evaluation for world models.arXiv preprint arXiv:2603.22212, 2026

    work page arXiv 2026

  25. [26]

    WorldLens: Full-Spectrum Evaluations of Driving World Models in Real World

    Ao Liang, Lingdong Kong, Tianyi Yan, Hongsi Liu, Wesley Yang, Ziqi Huang, Wei Yin, Jialong Zuo, Yixuan Hu, Dekai Zhu, et al. Worldlens: Full-spectrum evaluations of driving world models in real world.arXiv preprint arXiv:2512.10958, 2025. 12 WBENCH: A Benchmark for Interactive Video World Model EvaluationPREPRINT

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  26. [27]

    Video diffusion models

    Jonathan Ho, Tim Salimans, Alexey Gritsenko, William Chan, Mohammad Norouzi, and David J Fleet. Video diffusion models. Advances in neural information processing systems, 35:8633–8646, 2022

    2022

  27. [28]

    Align your latents: High-resolution video synthesis with latent diffusion models

    Andreas Blattmann, Robin Rombach, Huan Ling, Tim Dockhorn, Seung Wook Kim, Sanja Fidler, and Karsten Kreis. Align your latents: High-resolution video synthesis with latent diffusion models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 22563–22575, 2023

    2023

  28. [29]

    CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer

    Zhuoyi Yang, Jiayan Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, et al. Cogvideox: Text-to-video diffusion models with an expert transformer.arXiv preprint arXiv:2408.06072, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  29. [30]

    Sora 2.https://openai.com/zh-Hans-CN/index/sora-2/, 2025

    OpenAI. Sora 2.https://openai.com/zh-Hans-CN/index/sora-2/, 2025

    2025

  30. [31]

    Kling 3.0 pro.https://klingai.com, 2025

    Kuaishou Technology. Kling 3.0 pro.https://klingai.com, 2025

    2025

  31. [32]

    Veo 3: State-of-the-art video generation with audio

    Google DeepMind. Veo 3: State-of-the-art video generation with audio. https://deepmind.google/models/veo/, 2025

    2025

  32. [33]

    Wan: Open and Advanced Large-Scale Video Generative Models

    Team Wan, Ang Wang, Baole Ai, Bin Wen, Chaojie Mao, Chen-Wei Xie, Di Chen, Feiwu Yu, Haiming Zhao, Jianxiao Yang, et al. Wan: Open and advanced large-scale video generative models.arXiv preprint arXiv:2503.20314, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  33. [34]

    Seedance 2.0: Advancing Video Generation for World Complexity

    Team Seedance, De Chen, Liyang Chen, Xin Chen, Ying Chen, Zhuo Chen, Zhuowei Chen, Feng Cheng, Tianheng Cheng, Yufeng Cheng, et al. Seedance 2.0: Advancing video generation for world complexity.arXiv preprint arXiv:2604.14148, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  34. [35]

    Vidu q3 pro.https://www.vidu.com, 2025

    Shengshu Technology. Vidu q3 pro.https://www.vidu.com, 2025

    2025

  35. [36]

    LTX-Video: Realtime Video Latent Diffusion

    Yoav HaCohen, Nisan Chiprut, Benny Brazowski, Daniel Shalem, Dudu Moshe, Eitan Richardson, Eran Levin, Guy Shiran, Nir Zabari, Ori Gordon, et al. Ltx-video: Realtime video latent diffusion.arXiv preprint arXiv:2501.00103, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  36. [37]

    Cosmos world foundation models for physical ai

    Jinwei Gu. Cosmos world foundation models for physical ai. InProceedings of the 3rd International Workshop on Rich Media With Generative AI, pages 39–39, 2025

    2025

  37. [38]

    Longcat-video technical report.arXiv preprint arXiv:2510.22200, 2025

    Meituan LongCat Team, Xunliang Cai, Qilong Huang, Zhuoliang Kang, Hongyu Li, Shijun Liang, Liya Ma, Siyu Ren, Xiaoming Wei, Rixu Xie, et al. Longcat-video technical report.arXiv preprint arXiv:2510.22200, 2025

    work page arXiv 2025

  38. [39]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium.Advances in neural information processing systems, 30, 2017

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilibrium.Advances in neural information processing systems, 30, 2017

    2017

  39. [40]

    Towards Accurate Generative Models of Video: A New Metric & Challenges

    Thomas Unterthiner, Sjoerd Van Steenkiste, Karol Kurach, Raphael Marinier, Marcin Michalski, and Sylvain Gelly. Towards accurate generative models of video: A new metric & challenges.arXiv preprint arXiv:1812.01717, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  40. [41]

    World Models

    David Ha and Jürgen Schmidhuber. World models.arXiv preprint arXiv:1803.10122, 2(3):440, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  41. [42]

    A path towards autonomous machine intelligence version 0.9

    Yann LeCun et al. A path towards autonomous machine intelligence version 0.9. 2, 2022-06-27.Open Review, 62(1):1–62, 2022

    2022

  42. [43]

    Happy Oyster: An open-ended world model for real-time world creation and interaction

    Alibaba Token Hub. Happy Oyster: An open-ended world model for real-time world creation and interaction. https: //happyoyster.cn/, 2026

    2026

  43. [44]

    Marble: A multimodal world model.https://www.worldlabs.ai/blog/marble-world-model, 2025

    World Labs. Marble: A multimodal world model.https://www.worldlabs.ai/blog/marble-world-model, 2025

    2025

  44. [45]

    Vbench++: Comprehensive and versatile benchmark suite for video generative models.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    Ziqi Huang, Fan Zhang, Xiaojie Xu, Yinan He, Jiashuo Yu, Ziyue Dong, Qianli Ma, Nattapol Chanpaisit, Chenyang Si, Yuming Jiang, et al. Vbench++: Comprehensive and versatile benchmark suite for video generative models.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    2025

  45. [46]

    Evalcrafter: Benchmarking and evaluating large video generation models

    Yaofang Liu, Xiaodong Cun, Xuebo Liu, Xintao Wang, Yong Zhang, Haoxin Chen, Yang Liu, Tieyong Zeng, Raymond Chan, and Ying Shan. Evalcrafter: Benchmarking and evaluating large video generation models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 22139–22149, 2024

    2024

  46. [47]

    VideoPhy: Evaluating Physical Commonsense for Video Generation

    Hritik Bansal, Zongyu Lin, Tianyi Xie, Zeshun Zong, Michal Yarom, Yonatan Bitton, Chenfanfu Jiang, Yizhou Sun, Kai-Wei Chang, and Aditya Grover. Videophy: Evaluating physical commonsense for video generation.arXiv preprint arXiv:2406.03520, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  47. [48]

    Towards World Simulator: Crafting Physical Commonsense-Based Benchmark for Video Generation

    Fanqing Meng, Jiaqi Liao, Xinyu Tan, Wenqi Shao, Quanfeng Lu, Kaipeng Zhang, Yu Cheng, Dianqi Li, Yu Qiao, and Ping Luo. Towards world simulator: Crafting physical commonsense-based benchmark for video generation.arXiv preprint arXiv:2410.05363, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  48. [49]

    Worldscore: A unified evaluation benchmark for world generation

    Haoyi Duan, Hong-Xing Yu, Sirui Chen, Li Fei-Fei, and Jiajun Wu. Worldscore: A unified evaluation benchmark for world generation. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 27713–27724, 2025

    2025

  49. [50]

    Worldmodelbench: Judging video generation models as world models.arXiv preprint arXiv:2502.20694, 2025

    Dacheng Li, Yunhao Fang, Yukang Chen, Shuo Yang, Shiyi Cao, Justin Wong, Michael Luo, Xiaolong Wang, Hongxu Yin, Joseph E Gonzalez, et al. Worldmodelbench: Judging video generation models as world models.arXiv preprint arXiv:2502.20694, 2025

    work page arXiv 2025

  50. [51]

    Worldarena: A unified benchmark for evaluating perception and functional utility of embodied world models.arXiv preprint arXiv:2602.08971, 2026

    Yu Shang, Zhuohang Li, Yiding Ma, Weikang Su, Xin Jin, Ziyou Wang, Lei Jin, Xin Zhang, Yinzhou Tang, Haisheng Su, et al. Worldarena: A unified benchmark for evaluating perception and functional utility of embodied world models.arXiv preprint arXiv:2602.08971, 2026

    work page arXiv 2026

  51. [52]

    Video-bench: Human-aligned video generation benchmark

    Hui Han, Siyuan Li, Jiaqi Chen, Yiwen Yuan, Yuling Wu, Yufan Deng, Chak Tou Leong, Hanwen Du, Junchen Fu, Youhua Li, et al. Video-bench: Human-aligned video generation benchmark. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 18858–18868, 2025. 13 WBENCH: A Benchmark for Interactive Video World Model EvaluationPREPRINT

    2025

  52. [53]

    Fetv: A benchmark for fine-grained evaluation of open-domain text-to-video generation.Advances in Neural Information Processing Systems, 36: 62352–62387, 2023

    Yuanxin Liu, Lei Li, Shuhuai Ren, Rundong Gao, Shicheng Li, Sishuo Chen, Xu Sun, and Lu Hou. Fetv: A benchmark for fine-grained evaluation of open-domain text-to-video generation.Advances in Neural Information Processing Systems, 36: 62352–62387, 2023

    2023

  53. [54]

    "PhyWorldBench": A Comprehensive Evaluation of Physical Realism in Text-to-Video Models

    Jing Gu, Xian Liu, Yu Zeng, Ashwin Nagarajan, Fangrui Zhu, Daniel Hong, Yue Fan, Qianqi Yan, Kaiwen Zhou, Ming-Yu Liu, and Xin Eric Wang. "phyworldbench": A comprehensive evaluation of physical realism in text-to-video models, 2026. URL https://arxiv.org/abs/2507.13428

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  54. [55]

    Mmgr: Multi-modal generative reasoning.arXiv preprint arXiv:2512.14691, 2025

    Zefan Cai, Haoyi Qiu, Tianyi Ma, Haozhe Zhao, Gengze Zhou, Kung-Hsiang Huang, Parisa Kordjamshidi, Minjia Zhang, Wen Xiao, Jiuxiang Gu, et al. Mmgr: Multi-modal generative reasoning.arXiv preprint arXiv:2512.14691, 2025

    work page arXiv 2025

  55. [56]

    Worldbench: Disambiguating physics for diagnostic evaluation of world models.arXiv preprint arXiv:2601.21282, 2026

    Rishi Upadhyay, Howard Zhang, Jim Solomon, Ayush Agrawal, Pranay Boreddy, Shruti Satya Narayana, Yunhao Ba, Alex Wong, Celso M de Melo, and Achuta Kadambi. Worldbench: Disambiguating physics for diagnostic evaluation of world models.arXiv preprint arXiv:2601.21282, 2026

    work page arXiv 2026

  56. [57]

    4dworldbench: A comprehensive evaluation framework for 3d/4d world generation models.arXiv preprint arXiv:2511.19836, 2025

    Yiting Lu, Wei Luo, Peiyan Tu, Haoran Li, Hanxin Zhu, Zihao Yu, Xingrui Wang, Xinyi Chen, Xinge Peng, Xin Li, et al. 4dworldbench: A comprehensive evaluation framework for 3d/4d world generation models.arXiv preprint arXiv:2511.19836, 2025

    work page arXiv 2025

  57. [58]

    World reasoning arena, 2026

    PAN Team, Qiyue Gao, Kun Zhou, Jiannan Xiang, Zihan Liu, Dequan Yang, Junrong Chen, Arif Ahmad, Cong Zeng, Ganesh Bannur, Xinqi Huang, Zheqi Liu, Yi Gu, Yichi Yang, Guangyi Liu, Zhiting Hu, Zhengzhong Liu, and Eric Xing. World reasoning arena, 2026

    2026

  58. [59]

    Worldsimbench: Towards video generation models as world simulators.arXiv preprint arXiv:2410.18072, 2024

    Yiran Qin, Zhelun Shi, Jiwen Yu, Xijun Wang, Enshen Zhou, Lijun Li, Zhenfei Yin, Xihui Liu, Lu Sheng, Jing Shao, et al. Worldsimbench: Towards video generation models as world simulators.arXiv preprint arXiv:2410.18072, 2024

    work page arXiv 2024

  59. [60]

    Ewmbench: Evaluating scene, motion, and semantic quality in embodied world models.arXiv preprint arXiv:2505.09694, 2025

    Hu Yue, Siyuan Huang, Yue Liao, Shengcong Chen, Pengfei Zhou, Liliang Chen, Maoqing Yao, and Guanghui Ren. Ewmbench: Evaluating scene, motion, and semantic quality in embodied world models.arXiv preprint arXiv:2505.09694, 2025

    work page arXiv 2025

  60. [61]

    World-in-world: World models in a closed-loop world.arXiv preprint arXiv:2510.18135, 2025

    Jiahan Zhang, Muqing Jiang, Nanru Dai, Taiming Lu, Arda Uzunoglu, Shunchi Zhang, Yana Wei, Jiahao Wang, Vishal M Patel, Paul Pu Liang, et al. World-in-world: World models in a closed-loop world.arXiv preprint arXiv:2510.18135, 2025

    work page arXiv 2025

  61. [62]

    Drivinggen: A comprehensive benchmark for generative video world models in autonomous driving.arXiv preprint arXiv:2601.01528, 2026

    Yang Zhou, Hao Shao, Letian Wang, Zhuofan Zong, Hongsheng Li, and Steven L Waslander. Drivinggen: A comprehensive benchmark for generative video world models in autonomous driving.arXiv preprint arXiv:2601.01528, 2026

    work page arXiv 2026

  62. [63]

    Nano banana 2.https://blog.google/innovation-and-ai/technology/ai/nano-banana-2/, 2025

    Google. Nano banana 2.https://blog.google/innovation-and-ai/technology/ai/nano-banana-2/, 2025

    2025

  63. [64]

    GPT-Image-1.5.https://openai.com/zh-Hans-CN/index/new-chatgpt-images-is-here/, 2025

    OpenAI. GPT-Image-1.5.https://openai.com/zh-Hans-CN/index/new-chatgpt-images-is-here/, 2025

    2025

  64. [65]

    arXiv preprint arXiv:2508.03789 (2025) 4, 5, 10

    Yuhang Ma, Xiaoshi Wu, Keqiang Sun, and Hongsheng Li. Hpsv3: Towards wide-spectrum human preference score, 2025. URLhttps://arxiv.org/abs/2508.03789

    work page arXiv 2025

  65. [66]

    Megasam: Accurate, fast and robust structure and motion from casual dynamic videos

    Zhengqi Li, Richard Tucker, Forrester Cole, Qianqian Wang, Linyi Jin, Vickie Ye, Angjoo Kanazawa, Aleksander Holynski, and Noah Snavely. Megasam: Accurate, fast and robust structure and motion from casual dynamic videos. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10486–10496, 2025

    2025

  66. [67]

    DreamSim: Learning New Dimensions of Human Visual Similarity using Synthetic Data

    Stephanie Fu, Netanel Tamir, Shobhita Sundaram, Lucy Chai, Richard Zhang, Tali Dekel, and Phillip Isola. Dreamsim: Learning new dimensions of human visual similarity using synthetic data.arXiv preprint arXiv:2306.09344, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  67. [68]

    Transnet v2: An effective deep network architecture for fast shot transition detection

    Tomás Soucek and Jakub Lokoc. Transnet v2: An effective deep network architecture for fast shot transition detection. In Proceedings of the 32nd ACM International Conference on Multimedia, pages 11218–11221, 2024

    2024

  68. [69]

    SAM 2: Segment Anything in Images and Videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman Rädle, Chloe Rolland, Laura Gustafson, et al. Sam 2: Segment anything in images and videos.arXiv preprint arXiv:2408.00714, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  69. [70]

    Depth Anything 3: Recovering the Visual Space from Any Views

    Haotong Lin, Sili Chen, Junhao Liew, Donny Y Chen, Zhenyu Li, Guang Shi, Jiashi Feng, and Bingyi Kang. Depth anything 3: Recovering the visual space from any views.arXiv preprint arXiv:2511.10647, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  70. [71]

    Vggrpo: Towards world-consistent video generation with 4d latent reward.arXiv preprint arXiv:2603.26599, 2026

    Zhaochong An, Orest Kupyn, Théo Uscidda, Andrea Colaco, Karan Ahuja, Serge Belongie, Mar Gonzalez-Franco, and Marta Tintore Gazulla. Vggrpo: Towards world-consistent video generation with 4d latent reward.arXiv preprint arXiv:2603.26599, 2026

    work page arXiv 2026

  71. [72]

    VideoGPA: Distilling Geometry Priors for 3D-Consistent Video Generation

    Hongyang Du, Junjie Ye, Xiaoyan Cong, Runhao Li, Jingcheng Ni, Aman Agarwal, Zeqi Zhou, Zekun Li, Randall Balestriero, and Yue Wang. Videogpa: Distilling geometry priors for 3d-consistent video generation.arXiv preprint arXiv:2601.23286, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  72. [73]

    DINOv2: Learning Robust Visual Features without Supervision

    Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision.arXiv preprint arXiv:2304.07193, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  73. [74]

    Qwen3-VL Technical Report

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, Lianghao Deng, Wei Ding, Chang Gao, Chunjiang Ge, et al. Qwen3-vl technical report.arXiv preprint arXiv:2511.21631, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  74. [75]

    Seedance 1.5 pro: A Native Audio-Visual Joint Generation Foundation Model

    Team Seedance, Heyi Chen, Siyan Chen, Xin Chen, Yanfei Chen, Ying Chen, Zhuo Chen, Feng Cheng, Tianheng Cheng, Xinqi Cheng, et al. Seedance 1.5 pro: A native audio-visual joint generation foundation model.arXiv preprint arXiv:2512.13507, 2025. 14 WBENCH: A Benchmark for Interactive Video World Model EvaluationPREPRINT

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  75. [76]

    Videophy-2: A challenging action-centric physical commonsense evaluation in video generation.arXiv preprint arXiv:2503.06800, 2025

    Hritik Bansal, Clark Peng, Yonatan Bitton, Roman Goldenberg, Aditya Grover, and Kai-Wei Chang. Videophy-2: A challenging action-centric physical commonsense evaluation in video generation.arXiv preprint arXiv:2503.06800, 2025

    work page arXiv 2025

  76. [77]

    Kairos 3.0-4b: Real-time generative world model for embodied intelligence

    ACE Robotics. Kairos 3.0-4b: Real-time generative world model for embodied intelligence. https://github.com/ kairos-agi/kairos-sensenova/tree/main, 2026

    2026

  77. [78]

    Fantasyworld: Geometry-consistent world modeling via unified video and 3d prediction.arXiv preprint arXiv:2509.21657, 2025

    Yixiang Dai, Fan Jiang, Chiyu Wang, Mu Xu, and Yonggang Qi. Fantasyworld: Geometry-consistent world modeling via unified video and 3d prediction.arXiv preprint arXiv:2509.21657, 2025

    work page arXiv 2025

  78. [79]

    INSPATIO-WORLD: A Real-Time 4D World Simulator via Spatiotemporal Autoregressive Modeling

    InSpatio Team, Donghui Shen, Guofeng Zhang, Haomin Liu, Haoyu Ji, Hujun Bao, Hongjia Zhai, Jialin Liu, Jing Guo, Nan Wang, et al. Inspatio-world: A real-time 4d world simulator via spatiotemporal autoregressive modeling.arXiv preprint arXiv:2604.07209, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  79. [80]

    Astra: General interactive world model with autoregressive denoising.arXiv preprint arXiv:2512.08931, 2025

    Yixuan Zhu, Jiaqi Feng, Wenzhao Zheng, Yuan Gao, Xin Tao, Pengfei Wan, Jie Zhou, and Jiwen Lu. Astra: General interactive world model with autoregressive denoising.arXiv preprint arXiv:2512.08931, 2025

    work page arXiv 2025

  80. [81]

    Infinite-world: Scaling interactive world models to 1000-frame horizons via pose-free hierarchical memory

    Ruiqi Wu, Xuanhua He, Meng Cheng, Tianyu Yang, Yong Zhang, Zhuoliang Kang, Xunliang Cai, Xiaoming Wei, Chunle Guo, Chongyi Li, et al. Infinite-world: Scaling interactive world models to 1000-frame horizons via pose-free hierarchical memory. arXiv preprint arXiv:2602.02393, 2026

    work page arXiv 2026

Showing first 80 references.