Pith. sign in

REVIEW 4 major objections 7 minor 20 references

PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives

T0 review · 4 major / 7 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Agent players expose world models' long-horizon failures

desk verdict PlayWorld is a genuinely useful benchmark for interactive world models, but the central fairness claim about its adaptive agent-player protocol needs more evidence than the paper currently provides. read the letter →

arxiv 2608.13552 v1 pith:ZNXMFHBE submitted 2026-08-13 cs.CV

Kaixin Ding , Xi Chen , Minghong Cai , Zhiyuan Xu , Yiyang Wang , Yuxiang Lu , Junyi Li , Shuyang Chen
show 4 more authors
Yuan Gao Xin Tao Pengfei Wan Hengshuang Zhao
This is my paper · ORCID
classification cs.CV
keywords worldmodelsvideogenerationagentplayerlong-horizonevaluationinteractivebenchmarkgeometryconsistencystateevolutionVQArubric
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

PlayWorld is a benchmark that evaluates video world models the way a human player actually would: by giving each model a long-horizon objective, such as turning the camera 360 degrees or walking into water, and letting an AI agent adapt its actions online to achieve that objective. The paper argues that fixed action sequences, the standard evaluation method, produce incomparable scores across models because the same controls produce different effects in different models. Using 171 scenarios, a VQA rubric verifier, and nine state-of-the-art models, PlayWorld finds that current models remain unreliable on long-horizon interactive objectives, especially on maintaining spatial consistency and persistent state evolution.

What carries the argument

The key mechanism is the Agent Player, a closed-loop control system that observes a model's generated frames and action history and decides to keep, stop, extend, correct, or end the current action. This adapts execution to each model's action granularity, ensuring that all models are evaluated on the same objective rather than on identical—but possibly ineffective—action sequences. The scaffold also includes a VQA rubric verifier that answers sample-specific Yes/No questions after a trajectory-validity gate, converting answers into dimension scores on a 1–5 scale.

What would settle it

Run the PlayWorld protocol on a single model with two different agent models that have different action-adjustment styles (e.g., conservative vs. aggressive correction) and check whether the model's dimension scores change significantly; if they do, the benchmark is sensitive to agent behavior rather than isolating world-model ability.

Watch

Extended reading notes

Core claim

The central claim is that current video world models, though capable of short, visually plausible outputs, fail to sustain coherent worlds over long-horizon interaction. The paper demonstrates this failure through a benchmark that uses a multi-modal Agent Player to adapt actions toward a shared objective, and a VQA rubric that scores four capabilities: geometry consistency, interaction fidelity, out-of-sight evolution, and insight evolution. Across nine models, the scores are low, with out-of-sight evolution and insight evolution markedly worse, indicating a systemic inability to maintain persistent, causally consistent state changes over time.

Load-bearing premise

The Agent Player's online adaptation produces comparable execution of the same objective across models; if it compensates for one model more than another, the rubric scores may reflect the agent's behavior rather than the model's true ability.

Editorial extensions

If this is right

  • A benchmark with 171 scenarios and nine models showing low scores, especially in out-of-sight and insight evolution, suggests a concrete, measurable capability gap in current world models.
  • The approach invites model developers to target persistent state evolution explicitly, since current models fail to sustain changes during occlusion or long observation.
  • The finding that trajectory control can diverge from world-model ability implies that interaction fidelity and state evolution are distinct, separate axes of evaluation.
  • The paper's claim that fixed-action benchmarks conflate trajectory failure with world-model ability would, if true, invalidate a common evaluation practice in the field.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the VQA verifier aligns with human preference, a fully automated objective-driven benchmark could replace expensive human evaluation, but the robustness of the verifier across different objectives and models remains a key extension to test.
  • The finding that geometry consistency breaks down on revisitation suggests that world models lack a persistent global representation of the 3D world; training with explicit memory or 3D anchoring could be a direct extension.
  • The Agent Player's online adaptation might be more effective than a fixed trajectory, but it introduces the risk of confounding model ability with agent behavior; bounding this for the full nine-model set would be a logical next step.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. PlayWorld is a benchmark for evaluating interactive video world models under long-horizon, scene-grounded objectives. It comprises 171 human-annotated scenarios, each with an initial image, an objective such as 'rotate the camera 360 degrees' or 'walk into the water', a basic action sequence used as a shared reference, and a sample-specific VQA rubric. A multi-modal Agent Player (Claude Haiku 4.5) observes generated frames and adapts execution through Keep/Stop/Extend/Correct/End decisions; Gemini 3.1 Pro then scores each rollout across four dimensions (geometry consistency, interaction fidelity, out-of-sight evolution, insight evolution), after a trajectory-validity gate that assigns the minimum score of 1 to invalid rollouts. Nine world models are compared, with the headline finding that all models score poorly on out-of-sight and insight evolution (Table 2, maxima 1.81 and 1.95). The VQA scores are validated against human preferences (Spearman rho = 0.745-0.983, Appendix D), and basic ability metrics (VBench-style quality and VGGT-based controllability) are reported separately. The paper's central methodological claim is that closed-loop adaptation makes scores comparable across models with different action granularity, where fixed action-conditioned benchmarks conflate trajectory failure with world-model ability.

Significance. If the fairness protocol is substantiated, PlayWorld fills a real gap: existing benchmarks such as WorldScore, WorldMark, WBench, and MemoBench drive models with fixed low-level controls and cannot separate trajectory failure from world-model capability. The benchmark itself is substantial and carefully built: 171 human-curated cases, 820+ rubric questions, 1400+ videos, and a promise of code/data release. Several strengths deserve emphasis: the agent-model sensitivity analysis in Table 5 shows that the choice of agent model has limited effect on trajectory quality; Appendix B reports a second scoring pass with low aggregate variance (0.0112); the VQA scores correlate with human preferences on all four dimensions; and the headline finding, that all nine models are unreliable on long-horizon interactive objectives, is falsifiable and useful for the community. The circularity risk is low: rubrics are human-annotated, the agent and verifier are different models, and the human study is external to the rubric questions.

major comments (4)
  1. [§3.1, §4.1–4.2, Tables 2–4] The central claim that closed-loop adaptation yields comparable execution of the same objective across models (Sec. 3.1) is not yet substantiated by the reported evidence. Table 4 reports Agent-modified Action Ratios of 12.0% and 14.9% for only 25 cases and two models (Genie 3 and HappyOyster); no per-model decision statistics (Keep/Stop/Extend/Correct/End counts), per-case modification distributions, or modification rates for the other seven models are reported. This matters because Sec. 4.1 assigns the minimum score of 1 to any rollout that fails trajectory validation, and validation is applied to the agent-adapted rollout rather than to a common reference trajectory. The Table 2 dimension scores therefore mix rubric-measured capability, trajectory-control ability, and agent compensation in model-specific proportions. The geometry consistency pass rates in Table 3 range from 14.6% to 77.1%; for HY-WorldPlay, whose pass rate is 14.6%, the reported score 1.12 corresponds to a conditional rubric score of roughly 1.82 on the valid minority, so the composite is dominated by validation failure, whereas for Genie 3 the composite is mostly rubric-driven. The skeptic's concern is thus valid as stated. I ask for per-model adaptation statistics over the full benchmark, an analysis of whether validated rollouts reach comparable objective states (e.g., per-case completion-state or final-pose error), and a sensitivity analysis reporting rubric scores conditioned on validated rollouts alongside the gated composite.
  2. [§4.4, Fig. 6, Appendix D] The human-validation study establishes that the VQA verifier's ranking of the produced videos agrees with human pairwise preferences (Spearman rho = 0.933 Overall, 0.983 geometry, 0.933 interaction, 0.812 out-of-sight, 0.745 insight), but it does not bound the agent-compensation confound, because both the human raters and the verifier watched the same agent-adapted rollouts. Agreement between the two is exactly what one would expect even if the closed-loop controller had compensated unevenly across models, since both evaluate the same videos; the study therefore does not show that different models were tested at comparable objective states. The study is also modest in scale (120 pairwise items, 600 judgments, overall Fleiss' kappa = 0.434), and the weakest correlation occurs on insight evolution, the one dimension without a trajectory-validity gate. A control experiment that varies adaptation intensity on matched cases, or human judgments of whether paired rollouts from different models actually satisfy the stated completion condition, would speak to the protocol-level fairness claim that the current study does not address.
  3. [Appendix A, Tables 2–3] Two models are not evaluated on the full benchmark: HY-World2 skips cases whose initial images do not satisfy its input requirements, and those cases are excluded from its score aggregation, while Hunyuan-GameCraft-2 cannot always follow the intended WAIT-only protocol for insight evolution. These exclusions are disclosed honestly, but their impact on the cross-model comparison is not quantified, and Tables 2 and 3 present all nine models side by side as if computed over the same set of cases. I ask the authors to report the number of evaluated cases per model and per dimension and to recompute the Overall scores on the common subset of cases, so that ranking differences can be separated from case-coverage differences.
  4. [§3.4, Table 6] The Action Controllability pass rates inherit the same closed-loop concern: the target trajectory is 'defined by the executed actions, including online adjustments made by the Agent Player' (Sec. 3.4), so the target for each model is itself a product of the controller under test. A model whose commands were more heavily tuned through Extend/Stop/Correct decisions receives a target that is easier to match, and the pass rates are additionally computed only over rollouts with valid VGGT estimates. Because Sec. 4.3 draws conclusions about 'Action Controllability' from these numbers, the paper should report the per-model distribution of executed-action adjustments for the controllability subset, or separate 'the model follows the agent's commands' from 'the interaction reaches the intended objective state'.
minor comments (7)
  1. [§3.3] The Geometry consistency paragraph contains spelling artifacts ('TrajectoryV alidity', 'V alidity'), which appear to be spacing or formatting errors; the manuscript should be proofread for such issues.
  2. [Fig. 4] The caption of the interaction-fidelity example in Fig. 4 spells 'vender' instead of 'vendor'; please proofread the figure captions.
  3. [Fig. 1, Appendix A] The radar chart in Fig. 1 displays the four rubric dimensions on a 1–3 range and Basic Ability on a 0–1 range, as disclosed in Appendix A; the figure caption should state these ranges explicitly, since the chart's visual impression suggests a common scale.
  4. [§4.2] The sentence 'We attribute this limited sensitivity to the relatively constrained decision task...' is presented as an explanation without supporting evidence; either provide supporting measurements (such as the distribution of decision types or decision entropy) or reframe the sentence as a hypothesis.
  5. [Appendix B] The appendix appropriately cautions that the two-pass analysis 'is not a precise estimate of scoring uncertainty'; adding the per-question agreement rate between the two passes would be useful, since the aggregate mean variance of 0.0112 could hide systematic disagreements.
  6. [§3.1, §4.1] The relation between the 40-step interaction budget and the reported 'approximately 10 to 60 seconds' rollout duration is not explained; a short description of how steps, decision latency, and generated video length relate would clarify the protocol.
  7. [§3.4, Appendix A] Subject Consistency is reported as N/A when no dynamic subject is detected, and the controllability pass rates are computed over rollouts with valid estimates; the fraction of N/A or invalid cases per model should be reported so that readers can judge the coverage of the basic-ability metrics.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the benchmark's objectives, VQA rubrics, and human validation are independently constructed, and the adaptive-agent confound is an experimental-validity concern, not a derivation that reduces to its inputs.

full rationale

PlayWorld does not derive predictions from fitted parameters or from self-referential definitions. The benchmark content (171 scenarios, objectives, basic action sequences, and VQA rubrics) is human-annotated per Section 3.2, and the VQA verifier is an external LLM (Gemini 3.1 Pro) answering sample-specific questions, per Section 3.3 and Appendix B. The reported scores are aggregated from these verifier answers plus basic quality metrics, with no parameter fitted to the benchmark conclusions. Human validation in Section 4.4 provides an independent external check via 600 pairwise judgments. The only citation with author overlap is HyDRA (Chen et al., 2026c) used to adapt one video-quality metric, Subject Consistency, which is neither load-bearing for the central long-horizon evaluation nor a source of the main findings. The concern that the Agent Player's online adaptation (Keep/Stop/Extend/Correct/End) may confound cross-model comparison is an evaluation-validity limitation, not a circular derivation: the paper does not claim those scores follow by construction from the adaptation mechanism, and the confound is acknowledged indirectly by the limited 25-case ablation in Section 4.2. Thus the central comparison is self-contained and no circular step can be exhibited from the paper's equations or citations.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The benchmark introduces the Agent Player as a software component, not an unexplained physical entity. The load-bearing commitments are the hand-set scoring parameters and the reliability of the LLM verifier and agent, which are unchecked assumptions with partial validation.

free parameters (4)
  • Rubric question weights = regular questions w=1, defining question w=2
    Hand-set in Appendix A; directly controls dimension scores and therefore the model ranking.
  • Action pass-rate thresholds = translation error < 0.3, rotation error < 45 degrees
    Taken from WorldMark; no sensitivity analysis is provided for these thresholds in PlayWorld.
  • Contact-sheet sampling rates = 10 FPS with 5x5 grids; 0.5 FPS with 2x2 grids
    Manual choices for the VQA verifier input; they define what information the verifier can reason over.
  • Interaction budget = 40 steps
    Rollouts are capped at 40 agent-model interactions; this determines which long-horizon objectives can be completed.
assumptions (5)
  • domain assumption VQA responses of Gemini 3.1 Pro are valid for scoring world-model rollouts.
    All rubric scores in Tables 2 and 4 come from one LLM verifier; human validation covers 120 of over 1,400 videos and shows positive but not perfect correlation.
  • domain assumption The Agent Player reliably perceives frames and makes correct Keep/Stop/Extend/Correct/End decisions.
    Agent behavior is only ablated on 25 cases for two models, and per-model modification rates are not reported for the full benchmark.
  • domain assumption A shared basic action sequence plus agent adjustments preserves the same objective across models.
    This is the paper's fairness premise; the actual executed trajectory varies per model and the equivalence is not measured.
  • domain assumption Human-annotated objectives and rubrics are visually verifiable in generated rollouts.
    Annotators curate cases, but for out-of-sight and insight evolution the required state may be ambiguous in generated video.
  • domain assumption Standard vision tools (VGGT, Depth Anything V2, YOLO, CLIP) transfer to generated video.
    Used for pose estimation, depth stability, and subject consistency without validating on generated content.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives." pith.science (2026). https://pith.science/paper/ZNXMFHBE

@misc{pith2026260813552,
  author       = {Pith},
  title        = {Pith review of: PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZNXMFHBE}},
  note         = {Machine review of arXiv:2608.13552}
}
read the original abstract

Video world models simulate future states conditioned on current observations and user actions. Recent systems have demonstrated impressive video consistency and action controllability over long sequences. However, fairly comparing these interactive models remains challenging. In practice, a human player typically evaluates a world model by pursuing long-horizon objectives through interaction. For example, a user may turn around 360 degrees to see whether the environment remains consistent, or walk into the water and inspect whether realistic water ripples are generated. The action sequence required to achieve the same objective may vary substantially between models, making fixed action-conditioned evaluation unsuitable for cross-model comparison. To address this, we employ multi-modal Agent Players to interact with world models toward specified long-horizon objectives. Building on this paradigm, we introduce PlayWorld, a benchmark providing 171 scenarios, each with a specified objective. To evaluate performance thoroughly, we assess models along four core dimensions: geometry consistency, interaction fidelity, out-of-sight evolution, and insight evolution. In addition, we incorporate basic ability metrics for video quality and controllability. Experiments across nine state-of-the-art world models reveal that current models remain unreliable on long-horizon interactive objectives, particularly in maintaining spatial consistency and persistent state evolution. Code and data are available at https://github.com/kxding/PlayWorld.

Figures

Figures reproduced from arXiv: 2608.13552 by the authors.

Figure 1
Figure 1. PlayWorld evaluates world models from the perspective of a human player. We define long-horizon objectives such as “turning around 360 degrees” or “walking into the water”, and use an Agent Player to interact with the world models. We then evaluate them from multiple perspectives, including geometry consistency, interaction fidelity, and persistent state evolution. ABSTRACT Video world models simulate future states … view at source ↗
Figure 2
Figure 2. Benchmark construction pipeline of PlayWorld. Starting from diverse initial worlds, annotators define long-horizon objectives based on scenario-specific attributes, write basic action sequences, and construct sample-specific VQA rubrics. The resulting benchmark spans over 170 human-annotated cases and 50 action patterns over 10–60-second rollouts, with more than 820 questions applied to over 1,400 interactive videos… view at source ↗
Figure 3
Figure 3. Composition of PlayWorld. The benchmark comprises over 170 human-annotated cases across diverse scenario categories. The cases span four evaluation dimensions and diverse subject types, with dimension-specific VQA questions tailored to their evaluation criteria. Google Images (Google, 2026b), covering natural, urban, and fantasy environments with various subjects such as humans, animals, and vehicles. Annotators the… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Representative cases in PlayWorld. Each case combines an initial world, a scene￾grounded long-horizon objective, and a sample-specific VQA rubric. The expected trajectory is visualized alongside each case; for insight evolution, the camera remains stationary and observ…
Figure 5
Figure 5. Figure 5: Representative failures across four evaluation dimensions. Current video world models can follow camera or action controls, but still struggle to simulate coherent and realistic world dynamics. 4.2 AGENT PLAYER ANALYSIS Control strategy. Existing benchmarks typically e…
Figure 6
Figure 6. Figure 6: Alignment between human preferences and VQA metrics. [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: Agent interface execution on a web-based world model. (a) The agent interface enters the initial-world condition; (b) waits until the generated world becomes interactive; (c)–(d) dispatches controls and captures observations for the Agent Player’s online decisions; and…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 8 canonical work pages

  1. [4]

    GAIA-1: A generative world model for autonomous driving.arXiv preprint arXiv:2309.17080,

    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,

    arXiv

  2. [6]

    Yiran Qin, Zhelun Shi, Jiwen Yu, Xijun Wang, Enshen Zhou, Lijun Li, Zhenfei Yin, Xihui Liu, Lu Sheng, Jing Shao, Lei Bai, Wanli Ouyang, and Ruimao Zhang

    Accessed: 2026-05-03. Yiran Qin, Zhelun Shi, Jiwen Yu, Xijun Wang, Enshen Zhou, Lijun Li, Zhenfei Yin, Xihui Liu, Lu Sheng, Jing Shao, Lei Bai, Wanli Ouyang, and Ruimao Zhang. WorldSimBench: Towards Video Generation Models as World Simulators, October

    work page 2026

  3. [7]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision. In Marina Meila and Tong Zhang (eds.),Proceedings of the 38th International Conference on Machine Le...

    work page 2021

  4. [10]

    RoboScape: Physics-informed embodied world model.arXiv preprint arXiv:2506.23135,

    Yu Shang, Xin Zhang, Yinzhou Tang, Lei Jin, Chen Gao, Wei Wu, and Yong Li. RoboScape: Physics-informed embodied world model.arXiv preprint arXiv:2506.23135,

    arXiv

  5. [12]

    Tencent Hunyuan

    URL https://arxiv.org/ab s/2604.14268. Tencent Hunyuan. HY-World 1.5: A systematic framework for interactive world modeling with real-time latency and geometric consistency. https://3d-models.hunyuan.tencent. com/world/world1_5/HYWorld_1.5_Tech_Report.pdf,

    arXiv

  6. [13]

    VGGT: visual geometry grounded transformer

    Jianyuan Wang, Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rupprecht, and David Novotný. VGGT: visual geometry grounded transformer. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025, pp. 5294–

    work page 2025

  7. [14]

    URL https://openaccess.thecvf.com/content/CVPR2025/html/Wang_VGGT_Vi sual_Geometry_Grounded_Transformer_CVPR_2025_paper.html

    doi: 10.1109/CVPR52734.2025.00499. URL https://openaccess.thecvf.com/content/CVPR2025/html/Wang_VGGT_Vi sual_Geometry_Grounded_Transformer_CVPR_2025_paper.html. Xiaofeng Wang, Zheng Zhu, Guan Huang, Xinze Chen, Jiagang Zhu, and Jiwen Lu. DriveDreamer: Towards real-world-driven world models for autonomous driving.arXiv preprint arXiv:2309.09777,

    arXiv 2025

  8. [15]

    WorldRoamBench: An Open-World Benchmark for Long-Horizon Stability of Interactive World Models

    Ting-Bing Xu, Jiacheng Sui, Zhe Gao, Kewei Shi, Wenjin Yang, Zhicheng Liu, Zhaoxu Sun, Mingchao Sun, Hongyu Pan, Fan Jiang, Mu Xu, Qi Fan, Yang Gao, Yong Li, and Baoquan Chen. WorldRoamBench: An open-world benchmark for long-horizon stability of interactive world models.arXiv preprint arXiv:2606.31672, 2026a. Xiaojie Xu, Zhengyuan Lin, Kang He, Yukang Fen...

    work page Pith review arXiv

Show all 20 references
  1. [16]

    Yixuan Ye, Xuanyu Lu, Yuxin Jiang, Yuchao Gu, Rui Zhao, Qiwei Liang, Jiachun Pan, Fengda Zhang, Weijia Wu, and Alex Jinpeng Wang

    doi: 10.52202/079017-0688. 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, February

    doi

  2. [17]

    WBench: A comprehensive multi-turn benchmark for interactive video world model evaluation.arXiv preprint arXiv:2605.25874,

    Kaining Ying, Hengrui Hu, Siyu Ren, Jiamu Li, Fengjiao Chen, Ziwen Wang, Xuezhi Cao, Xunliang Cai, and Henghui Ding. WBench: A comprehensive multi-turn benchmark for interactive video world model evaluation.arXiv preprint arXiv:2605.25874,

    arXiv

  3. [18]

    IRASim: A fine-grained world model for robot manipulation.arXiv preprint arXiv:2406.14540,

    Fangqi Zhu, Hongtao Wu, Song Guo, Yuxiao Liu, Chilam Cheang, and Tao Kong. IRASim: A fine-grained world model for robot manipulation.arXiv preprint arXiv:2406.14540,

    arXiv

  4. [19]

    SANA-WM: Efficient minute-scale world modeling with hybrid linear diffusion transformer, 2026a

    Haoyi Zhu, Haozhe Liu, Yuyang Zhao, Tian Ye, Junsong Chen, Jincheng Yu, Tong He, Song Han, and Enze Xie. SANA-WM: Efficient minute-scale world modeling with hybrid linear diffusion transformer, 2026a. URLhttps://arxiv.org/abs/2605.15178. Yixuan Zhu, Jiaqi Feng, Wenzhao Zheng, ...

    arXiv

  5. [20]

    as the agent model for all reported benchmark experiments. Model execution.Genie 3, LingBot-World, LingBot-World2, HY-World2, and HappyOyster are evaluated through their user-facing web interfaces, whereas SANA-WM, Hunyuan-GameCraft-2, HY-WorldPlay, and Matrix-Game-3.0 are eva...

    2024

  6. [21]

    Subject Consistency detects dynamic subjects with YOLO (Redmon et al., 2016), extracts CLIP (Radford et al.,

    following MemoBench (Chen et al., 2026b); and Subject Consistency is adapted from HyDRA’sDSCctx (Chen et al., 2026c). Subject Consistency detects dynamic subjects with YOLO (Redmon et al., 2016), extracts CLIP (Radford et al.,

    2016

  7. [2016]

    URL https://doi.org/10.1109/CVPR.2 016.91

    doi: 10.1109/CVPR.2016.91. URL https://doi.org/10.1109/CVPR.2 016.91. Robbyant Team, Zelin Gao, Qiuyu Wang, Yanhong Zeng, Jiapeng Zhu, Ka Leong Cheng, Yixuan Li, Hanlin Wang, Yinghao Xu, Shuailei Ma, Yihang Chen, Jie Liu, Yansong Cheng, Yao Yao, Jiayi Zhu, Yihao Meng, Kecheng ...

    2016 doi

  8. [2021]

    Joseph Redmon, Santosh Kumar Divvala, Ross B

    URLhttp://proceedings.mlr.press/v139/radford21a.html. Joseph Redmon, Santosh Kumar Divvala, Ross B. Girshick, and Ali Farhadi. You only look once: Unified, real-time object detection. In2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las V egas, NV ...

    2016

  9. [2023]

    DrivingWorld: Constructing world model for autonomous driving via video GPT.arXiv preprint arXiv:2412.19505,

    Xiaotao Hu, Wei Yin, Mingkai Jia, Junyuan Deng, Xiaoyang Guo, Qian Zhang, Xiaoxiao Long, and Ping Tan. DrivingWorld: Constructing world model for autonomous driving via video GPT.arXiv preprint arXiv:2412.19505,

    arXiv

  10. [2024]

    Freeman, Jitendra Malik, Pieter Abbeel, Russ Tedrake, Vincent Sitzmann, and Yilun Du

    12 Preprint Boyuan Chen, Tianyuan Zhang, Haoran Geng, Caiyi Zhang, Peihao Li, Kiwhan Song, William T. Freeman, Jitendra Malik, Pieter Abbeel, Russ Tedrake, Vincent Sitzmann, and Yilun Du. Large Video Planner Enables Generalizable Robot Control, May 2026a. Haoyu Chen, Kaichen Z...

    arXiv

  11. [2025]

    Vista: A generalizable driving world model with high fidelity and versatile controllability.arXiv preprint arXiv:2405.17398,

    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.arXiv preprint arXiv:2405.17398,

    arXiv

  12. [2026]

    URLhttps://arxiv.org/abs/2607.07534. Google. Gemini 3.1 pro. https://ai.google.dev/gemini-api/docs/models, 2026a. Gemini API model documentation. Google. Google images.https://images.google.com/, 2026b. Accessed July

    arXiv

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.