Skip to content

About

πŸ”₯ Quo Vadis, World Modeling? Towards Interactive World Proxies for Continually Improving Agents

worldbench.github.io/awesome-agentic-world-model

Topics

agentagenticagentic-aiagentic-world-modelagentsaigcclosed-loopclosed-loop-controlclosed-loop-experimentsclosed-loop-simulationgenerationgenerative-aiimage-generationvideo-generationworld-modelworld-modelingworld-modelsworld-models-rl

Resources

Readme
MIT license
Activity
Custom properties

Stars

62 stars

Watchers

2 watching

Forks

1 fork
Report repository
BranchesTags

Latest commit

Β 

History

17 Commits

Folders and files

NameName
Last commit message
Last commit date
LICENSE
LICENSE
Β 
Β 
README.md
README.md
Β 
Β 
index.html
index.html
Β 
Β 

Repository files navigation

Awesome PR's Welcome Visitors arXiv Hugging Face Papers

Awesome Agentic World Modeling

This repository accompanies the position paper "Quo Vadis, World Modeling? Towards Interactive World Proxies for Continually Improving Agents," and curates the works it reviews. The survey argues for a shift from a World Model that passively predicts the next physical state toward an Agent-Centric World Proxy: an environment-grounded interface that returns the information transition an agent needs (a future state, rendered view, execution result, retrieved memory or skill, or verdict on a plan) to plan, learn, and continually improve.

The taxonomy has two orthogonal axes:

  • Six proxy functions (what feedback the proxy returns): Dynamics, Spatial, Execution, Memory / Experience, Skill, and Reward / Verification.
  • Three empowerment levels (how the proxy improves the agent): L1 inference-time guidance, L2 training-time optimization, and L3 Agent-Proxy co-evolution.

Call for Contributions

Agentic world modeling moves fast, and the boundary between a passive world model and an agent-facing world proxy is still being drawn. This list aims to be a living, community-maintained map of that space, and contributions of new work or corrections are very welcome.

To contribute, open an issue or a pull request:

  • Where: add each entry to the section matching its proxy function (or the closest one), and keep every table in chronological order.
  • Format: follow the existing row layout: a short model name in Model (or - if none), the title with an arXiv badge in Paper, and the abbreviated venue with year in Venue.
  • Verify: check the title, authors, arXiv ID, and venue against the primary source before adding, and prefer the formally published version whenever one exists.
  • Fix: corrections to venues, broken links, or mis-categorized entries are equally welcome; flag them in an issue or fix them directly in a pull request.

Table of Contents

Background

Classical World Models & Model-Based RL

Foundations: state-transition world models and model-based RL that predict the next physical state given a state and an action.

Model Paper Venue
- Dyna, an integrated architecture for learning, planning, and reacting SIGART 1991
- arXiv
Recurrent World Models Facilitate Policy Evolution		 NeurIPS 2018
- arXiv
When to Trust Your Model: Model-Based Policy Optimization		 NeurIPS 2019
- arXiv
Dream to Control: Learning Behaviors by Latent Imagination		 ICLR 2020
- arXiv
Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model		 Nature 2020
- A Path Towards Autonomous Machine Intelligence OpenReview preprint 2022
- arXiv
Model-Based Reinforcement Learning: A Survey		 FTML 2023
- arXiv
Mastering diverse control tasks through world models		 Nature 2025

Continual Improvement & Interaction Cost

Why real-environment interaction alone cannot carry continual improvement: cost, safety, latency, and the need for scalable feedback.

Model Paper Venue
MuJoCo MuJoCo: A Physics Engine for Model-Based Control IROS 2012
- arXiv
Isaac Gym: High Performance GPU-Based Physics Simulation for Robot Learning		 NeurIPS 2021
DayDreamer arXiv
DayDreamer: World Models for Physical Robot Learning		 CoRL 2023
CodeIt arXiv
CodeIt: Self-Improving Language Models with Prioritized Hindsight Replay		 ICML 2024
Voyager arXiv
Voyager: An Open-Ended Embodied Agent with Large Language Models		 TMLR 2024
OSWorld arXiv
OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments		 NeurIPS 2024
WebArena arXiv
WebArena: A Realistic Web Environment for Building Autonomous Agents		 ICLR 2024
CASCADE arXiv
CASCADE: Cumulative Agentic Skill Creation through Autonomous Development and Evolution		 arXiv 2025
- arXiv
Continual Learning and Catastrophic Forgetting		 Learning and Memory: A Comprehensive Reference 2025
- arXiv
Agent Learning via Early Experience		 ICML 2026

The Six Proxy Functions

Each proxy function answers a different question for the agent, but all return agent-usable feedback.

Dynamics Proxy

How would the environment change if the agent executed a given action? Video, driving, game, and model-based dynamics predictors.

Model Paper Venue
- Learning Latent Dynamics for Planning from Pixels ICML 2019
FitVid arXiv
FitVid: Overfitting in Pixel-Level Video Prediction		 arXiv 2021
VideoGPT arXiv
VideoGPT: Video Generation using VQ-VAE and Transformers		 arXiv 2021
- arXiv
Video Diffusion Models		 NeurIPS 2022
- arXiv
Diffusion Models for Video Prediction and Infilling		 TMLR 2022
MCVD arXiv
MCVD: Masked Conditional Video Diffusion for Prediction, Generation, and Interpolation		 NeurIPS 2022
GAIA-1 arXiv
GAIA-1: A Generative World Model for Autonomous Driving		 arXiv 2023
- arXiv
Transformers Are Sample-Efficient World Models		 ICLR 2023
STORM STORM: Efficient Stochastic Transformer based World Models for Reinforcement Learning NeurIPS 2023
- arXiv
Diffusion for World Modeling: Visual Details Matter in Atari		 NeurIPS 2024
- Video generation models as world simulators 2024
Genie arXiv
Genie: Generative Interactive Environments		 ICML 2024
Vista arXiv
Vista: A Generalizable Driving World Model with High Fidelity and Versatile Controllability		 NeurIPS 2024
TD-MPC2 arXiv
TD-MPC2: Scalable, Robust World Models for Continuous Control		 ICLR 2024
RoboCasa RoboCasa: Large-Scale Simulation of Everyday Tasks for Generalist Robots RSS 2024
DriveDreamer arXiv
DriveDreamer: Towards Real-World-Driven World Models for Autonomous Driving		 ECCV 2024
iVideoGPT iVideoGPT: Interactive VideoGPTs are Scalable World Models NeurIPS 2024
- arXiv
Learning Interactive Real-World Simulators		 ICLR 2024
DynamicCity arXiv
DynamicCity: Large-Scale 4D Occupancy Generation from Dynamic Scenes		 ICLR 2025
GameGen-X arXiv
GameGen-X: Interactive Open-world Game Video Generation		 ICLR 2025
- Genie 3: A New Frontier for World Models 2025
- arXiv
Towards Unified World Models for Visual Navigation via Memory-Augmented Planning and Foresight		 arXiv 2025
- arXiv
World Modelling Improves Language Model Agents		 arXiv 2025
MineWorld arXiv
MineWorld: a Real-Time and Open-Source Interactive World Model on Minecraft		 arXiv 2025
MobileWorldBench arXiv
MobileWorldBench: Towards Semantic World Modeling For Mobile Agents		 arXiv 2025
- World Model on Million-Length Video And Language With Blockwise RingAttention ICLR 2025
Vision arXiv
Vision-Centric 4D Occupancy Forecasting and Planning via Implicit Residual World Models		 arXiv 2025
- arXiv
V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning		 arXiv 2025
- Diffusion Models Are Real-Time Game Engines ICLR 2025
Marble Marble: A Multimodal World Model 2025
- arXiv
Driving in the occupancy world: Vision-centric 4d occupancy forecasting and planning via world models for autonomous driving		 AAAI 2025
Matrix-Game arXiv
Matrix-Game: Interactive World Foundation Model		 arXiv 2025
Language arXiv
Language-Conditioned World Modeling for Visual Navigation		 arXiv 2026
- arXiv
Advancing Open-source World Models		 arXiv 2026
NavThinker arXiv
NavThinker: Action-Conditioned World Models for Coupled Prediction and Planning in Social Navigation		 arXiv 2026
LiDARCrafter arXiv
LiDARCrafter: Dynamic 4D World Modeling from LiDAR Sequences		 AAAI 2026
Self arXiv
Self-Execution Simulation Improves Coding Models		 arXiv 2026
- arXiv
Is Your Driving World Model an All-Around Player?		 CVPR 2026
U4D arXiv
U4D: Uncertainty-Aware 4D World Modeling from LiDAR Sequences		 CVPR 2026
AD-R1 arXiv
AD-R1: Closed-Loop Reinforcement Learning for End-to-End Autonomous Driving with Impartial World Models		 CVPR 2026
SPIRAL arXiv
SPIRAL: Self-Evolving Action-Conditioned Video Generation via Reflective Planning Agents		 arXiv 2026
FlowWAM arXiv
FlowWAM: Optical Flow as a Unified Action Representation for World Action Models		 arXiv 2026
LeapBot arXiv
LeapBot-WA: World-Anchor Action Models via Predictive Latent Alignments		 arXiv 2026
Enfold arXiv
Enfold: Folding World Model Imagination into Predictive Representations for Ultra-Efficient Embodied Control		 arXiv 2026
- arXiv
World Action Planner: Generalizable Decision-Making with Action-Conditioned World Models		 arXiv 2026
Auto arXiv
Auto-JEPA: A Latent World Model of Continuous Intent for End-to-End Autonomous Driving		 arXiv 2026
Faster arXiv
Faster-WAM: Do World Action Models Need Deep Action Modules?		 arXiv 2026
DF$^3$ arXiv
DF$^3$: World Modeling via Decoder-Free Feature Forecasting in Autonomous Navigation		 arXiv 2026
SG arXiv
SG-WAM: Self-Guided World Modeling in Geometry-Aware Policy Space		 arXiv 2026
LAWM arXiv
LAWM-3D: Learning 3D-Aware Latent Actions from Human Videos for Generalizable Robot World Models		 arXiv 2026
XEWorld arXiv
XEWorld: Can Action-Conditioned World Models Generalize to Unseen Robot Embodiments?		 arXiv 2026
Adaptive arXiv
Adaptive-WAM: Quality-Guided Early-Exit Planning from Intermediate Video-Diffusion Features		 arXiv 2026
GWM arXiv
GWM-VLA: Geometry-Aware Latent World Modeling for Vision-Language-Action Learning		 arXiv 2026
4D arXiv
4D-WAM: Infusing Spatiotemporal Awareness into World Action Models through Trajectory Fields		 arXiv 2026
WA arXiv
WA-SpecDec: World-Aware Speculative Decoding for Vision-Language-Action Models		 arXiv 2026
JEPA arXiv
JEPA-WAM: Learning Vision-Language-Action Policies with Joint-Embedding World Modeling		 arXiv 2026
- arXiv
World Tokens: Enhancing Embodied Policies with Training-Time World Modeling		 arXiv 2026
SLIM arXiv
SLIM-0.5B: Learning Action-Grounded Predictive Latents for Robot Manipulation		 arXiv 2026
Energy arXiv
Energy-Structured Latent World Models with Neural Time Fields for Physically Constistent Open-World Motion Planning		 arXiv 2026
Dreamer arXiv
Dreamer-SAC: Off-Policy Learning in Latent World Models for Sample-Efficient Autonomous Driving		 arXiv 2026
BrainWAM arXiv
BrainWAM: Action-Space Coordination of Semantic Priors and Predictive Dynamics for Autonomous Driving		 arXiv 2026
DreamX arXiv
DreamX-Phi 1.0: Action-Conditioned Video World Model for Robotic Manipulation		 arXiv 2026
ForgeWM arXiv
ForgeWM: Progressive Causal Training for Few-Step Action-Conditioned Video World Models		 arXiv 2026
- arXiv
Q-Learning With World Models		 arXiv 2026
Hydra arXiv
Hydra-0: Action Flow for Generalist World Modeling and Control		 arXiv 2026
GigaBrain arXiv
GigaBrain-WBC-0.5: A Behavior World Model for Robust Whole-Body Control with Environment Interaction		 arXiv 2026
- arXiv
Reinforced Planning with Latent World Models		 arXiv 2026
DA arXiv
DA-WAM: Decision-Aligned Future Latents for Driving World Models		 arXiv 2026
- arXiv
Temporally Centered SIGReg Improves Multi-Task LeWorldModel Learning: From Analysis to Method		 arXiv 2026
- arXiv
Overcoming Statistical Bias in Action-Controllable World Models		 arXiv 2026
- arXiv
Mitigating Compounding Error via Video Representation Regularization		 arXiv 2026
ODEWorld arXiv
ODEWorld: A Continuous Predictive Architecture via Physical-Time Flow		 arXiv 2026
GeniWorld arXiv
GeniWorld: A Generalizable Interactive World Model for Robotic Manipulation via Visual Actions		 arXiv 2026
ContactFlow arXiv
ContactFlow: A video action conditioning that transfers across embodiments		 arXiv 2026
EmbodiedVAE arXiv
EmbodiedVAE: Disentangled Video VAE for Efficient and Controllable Embodied Manipulation		 arXiv 2026
- arXiv
$Ο„_0$-VLA: a Hierarchical Robot Foundation Model with World-Model-Guided Test-Time Computation		 arXiv 2026
Depth arXiv
Depth-Regularized JEPA World Models Learn More Transferable Representations from Real Outdoor Robot Data		 arXiv 2026
- arXiv
Is Forward Prediction Enough? Physical State Grounding for JEPA World Models		 arXiv 2026
Xiaomi arXiv
Xiaomi-Robotics-U0: Unified Embodied Synthesis with World Foundation Model		 arXiv 2026
MiniWorld arXiv
MiniWorld: Democratizing the Training of Video World Models from Scratch		 arXiv 2026
BWM arXiv
BWM: A Low-Cost High-Fidelity World Simulator for Robot Learning		 arXiv 2026
S2 arXiv
S2-HWM: Sparse Event-Structured Hierarchical World Model for Long-Horizon Surgical Robot Manipulation		 arXiv 2026
- arXiv
Streaming Multi-Agent Autoregressive Diffusion Model with World State Registers		 arXiv 2026
Population arXiv
Population-Scalable Multi-Agent World Modeling		 arXiv 2026
DreamQAS arXiv
DreamQAS: Learning a Decision-Useful World Model for VQE-Efficient Quantum Architecture Search		 arXiv 2026
FedWorld arXiv
FedWorld: Scope-Aware Federation of Agent World Models		 arXiv 2026
- arXiv
Offline Multi-Agent Reinforcement Learning with a Physics-Informed World Model for Cooperative Mixed Traffic Control		 arXiv 2026
DreamTrajectory arXiv
DreamTrajectory: Trajectory-Guided Action Generation with World Model Alignment for Mobile Manipulation		 arXiv 2026
EndoWAM arXiv
EndoWAM: A Grounded World-Action Model for Generalizable Endoscopic Navigation		 arXiv 2026
- arXiv
Surgical WAM: A World-Action Model for Data-Efficient Surgical Robot Learning		 arXiv 2026
Motus2 arXiv
Motus2: A Self-Evolving General World Model for Dexterous Manipulation		 arXiv 2026
LeWM+I arXiv
Keeping JEPA World Models Plannable When Little of the Frame Moves		 arXiv 2026
SufficientPlan arXiv
How Much Planning Is Enough? Reducing Search and Computation in World-Model Planning		 arXiv 2026
UNITAS arXiv
UNITAS: A 3D-Native World Action Model for Embodied Manipulation		 arXiv 2026
- arXiv
Back on Track: Deployment-time Policy Improvement via World Model Planning		 arXiv 2026
VQ-WM arXiv
Vector-Quantized World Model: Latent Feedback Motion Planning for Robot Manipulation		 arXiv 2026
ActTime arXiv
ActTime: Action-Conditioned World Modeling for Time Series		 arXiv 2026
EmbodiedWorld-200K arXiv
EmbodiedWorld-200K: Toward Embodied World Agents via Embodied-Planning Dataset and Interactive World Models Β· Project 		arXiv 2026
Ropedia W0 arXiv
Ropedia W0: Scaling Loco-Manipulation World Models across Embodiments with Human Experience Β· Project 		arXiv 2026

Spatial Proxy

What would the agent observe from another viewpoint or position? NeRF/Gaussian rendering, 3D/4D reconstruction and generation, and visual imagination for navigation.

Model Paper Venue
NeRF arXiv
NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis		 ECCV 2020
Mip-NeRF arXiv
Mip-NeRF: A Multiscale Representation for Anti-Aliasing Neural Radiance Fields		 ICCV 2021
PathDreamer arXiv
PathDreamer: A World Model for Indoor Navigation		 ICCV 2021
Mip arXiv
Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fields		 CVPR 2022
SceneDreamer arXiv
SceneDreamer: Unbounded 3D Scene Generation from 2D Image Collections		 TPAMI 2023
SceneScape arXiv
SceneScape: Text-Driven Consistent Scene Generation		 NeurIPS 2023
Text2Room arXiv
Text2Room: Extracting Textured 3D Meshes from 2D Text-to-Image Models		 ICCV 2023
VoxPoser arXiv
VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models		 CoRL 2023
- arXiv
3D Gaussian Splatting for Real-Time Radiance Field Rendering		 TOG 2023
Nerfstudio arXiv
Nerfstudio: A Modular Framework for Neural Radiance Field Development		 SIGGRAPH Asia 2023
- arXiv
A Survey on 3D Gaussian Splatting		 CSUR 2024
- arXiv
3D Gaussian as a New Era: A Survey		 TVCG 2024
- arXiv
Grounding Image Matching in 3D with MASt3R		 ECCV 2024
DreamScene arXiv
DreamScene: 3D Gaussian-based Text-to-3D Scene Generation via Formation Pattern Sampling		 ECCV 2024
Scaffold-GS arXiv
Scaffold-GS: Structured 3D Gaussians for View-Adaptive Rendering		 CVPR 2024
Dreamforge arXiv
Dreamforge: Motion-aware autoregressive video generation for multi-view driving scenes		 arXiv 2024
DriveWorld DriveWorld: 4D Pre-trained Scene Understanding via World Models for Autonomous Driving CVPR 2024
DimensionX arXiv
DimensionX: Create Any 3D and 4D Scenes from a Single Image with Controllable Video Diffusion		 arXiv 2024
ManiSkill3 arXiv
ManiSkill3: GPU Parallelized Robotics Simulation and Rendering for Generalizable Embodied AI		 arXiv 2024
DUSt3R arXiv
DUSt3R: Geometric 3D Vision Made Easy		 CVPR 2024
- arXiv
4D Gaussian Splatting for Real-Time Dynamic Scene Rendering		 CVPR 2024
CityDreamer arXiv
CityDreamer: Compositional Generative Model of Unbounded 3D Cities		 CVPR 2024
Mip-Splatting arXiv
Mip-Splatting: Alias-Free 3D Gaussian Splatting		 CVPR 2024
Text2NeRF arXiv
Text2NeRF: Text-Driven 3D Scene Generation with Neural Radiance Fields		 TVCG 2024
Copilot4D Copilot4D: Learning Unsupervised World Models for Autonomous Driving via Discrete Diffusion ICLR 2024
OccWorld arXiv
OccWorld: Learning a 3D Occupancy World Model for Autonomous Driving		 ECCV 2024
ReCamMaster arXiv
ReCamMaster: Camera-Controlled Generative Rendering from A Single Video		 arXiv 2025
- arXiv
Navigation world models		 CVPR 2025
PhysX-Anything arXiv
PhysX-Anything: Simulation-Ready Physical 3D Assets from Single Image		 arXiv 2025
GAF arXiv
GAF: Gaussian Action Field as a 4D Representation for Dynamic World Modeling in Robotic Manipulation		 arXiv 2025
ParticleFormer arXiv
ParticleFormer: A 3D Point Cloud World Model for Multi-Object, Multi-Material Robotic Manipulation		 arXiv 2025
- arXiv
3D and 4D World Modeling: A Survey		 arXiv 2025
GWM GWM: Towards Scalable Gaussian World Models for Robotic Manipulation ICCV 2025
MindCube arXiv
MindCube: Spatial Mental Modeling from Limited Views		 arXiv 2025
GAIA-2 arXiv
GAIA-2: A Controllable Multi-View Generative World Model for Autonomous Driving		 arXiv 2025
- arXiv
3D Reconstruction with Spatial Memory		 3DV 2025
- arXiv
Continuous 3D Perception Model with Persistent State		 arXiv 2025
Fast3R arXiv
Fast3R: Towards 3D Reconstruction of 1000+ Images in One Forward Pass		 arXiv 2025
WonderWorld arXiv
WonderWorld: Interactive 3D Scene Generation from a Single Image		 CVPR 2025
- arXiv
How to Enable LLM with 3D Capacity? A Survey of Spatial Reasoning in LLM		 IJCAI 2025
- arXiv
Occupancy World Model for Robots		 arXiv 2025
TesserAct arXiv
TesserAct: Learning 4D Embodied World Models		 arXiv 2025
Aether arXiv
Aether: Geometric-Aware Unified World Modeling		 ICCV 2025
OccuBench arXiv
OccuBench: Evaluating AI Agents on Real-World Professional Tasks via Language World Models		 arXiv 2026
PointWorld arXiv
PointWorld: Scaling 3D World Models for In-The-Wild Robotic Manipulation		 arXiv 2026
CounterScene arXiv
CounterScene: Counterfactual Causal Reasoning in Generative World Models for Safety-Critical Closed-Loop Evaluation		 arXiv 2026
MVISTA-4D arXiv
MVISTA-4D: View-Consistent 4D World Model with Test-Time Action Inference for Robotic Manipulation		 arXiv 2026
X-scene arXiv
X-scene: Large-scale driving scene generation with high fidelity and flexible controllability		 NeurIPS 2026
Code2World arXiv
Code2World: A GUI World Model via Renderable Code Generation		 arXiv 2026
UniNav arXiv
UniNav: A Unified World-Action Diffusion Model for Visual Navigation		 arXiv 2026
- arXiv
SC$^{2}$-WM: A Self-Correcting World Model with Closed-Loop Feedback for Vision-and-Language Navigation in Continuous Environments		 arXiv 2026
- arXiv
Latent World Models with Monotone Planning Costs for Image-Goal Navigation		 arXiv 2026
Orbit arXiv
Orbit-Planner: Towards Latent World Models for On-Orbit Obstacle Avoidance of Satellite Agents		 arXiv 2026
World arXiv
World-Model-Grounded LLM Planning for AUV and ASV Navigation Near Offshore Wind Farms		 arXiv 2026

Execution Proxy

What would the digital environment return after an operation? Web, GUI, code, shell, and API/tool execution predictors.

Model Paper Venue
RT-2 arXiv
RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control		 CoRL 2023
- arXiv
Code as Policies: Language Model Programs for Embodied Control		 ICRA 2023
Toolformer arXiv
Toolformer: Language Models Can Teach Themselves to Use Tools		 NeurIPS 2023
- arXiv
Generating Code World Models with Large Language Models Guided by Monte Carlo Tree Search		 NeurIPS 2024
AgentBench arXiv
AgentBench: Evaluating LLMs as Agents		 ICLR 2024
- arXiv
WorldCoder, a Model-Based LLM Agent: Building World Models by Writing Code and Interacting with the Environment		 NeurIPS 2024
Math-Shepherd arXiv
Math-Shepherd: Verify and Reinforce LLMs Step-by-Step without Human Annotations		 ACL 2024
- arXiv
Web Agents with World Models: Learning and Leveraging Environment Dynamics in Web Navigation		 ICLR 2025
CWM arXiv
CWM: An Open-Weights LLM for Research on Code Generation with World Models		 arXiv 2025
- arXiv
Web World Models		 arXiv 2025
WebSynthesis arXiv
WebSynthesis: World-Model-Guided MCTS for Efficient WebUI-Trajectory Synthesis		 arXiv 2025
- arXiv
Is Your LLM Secretly a World Model of the Internet? Model-Based Planning for Web Agents		 TMLR 2025
MegaSaM arXiv
MegaSaM: Accurate, Fast, and Robust Structure and Motion from Casual Dynamic Videos		 CVPR 2025
ToolSandbox arXiv
ToolSandbox: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities		 NAACL 2025
ViMo arXiv
ViMo: A Generative Visual GUI World Model for App Agents		 arXiv 2025
- arXiv
Cosmos World Foundation Model Platform for Physical AI		 arXiv 2025
GTM arXiv
GTM: Simulating the World of Tools for AI Agents		 arXiv 2025
- arXiv
The Role of World Models in Shaping Autonomous Driving: A Comprehensive Survey		 arXiv 2025
GameFactory GameFactory: Creating New Games with Generative Interactive Videos ICCV 2025
ComAct arXiv
ComAct: Reframing Professional Software Manipulation via COM-as-Action Paradigm		 arXiv 2026
MobileDreamer arXiv
MobileDreamer: Generative Sketch World Model for GUI Agent		 arXiv 2026
Computer arXiv
Computer-Using World Model		 arXiv 2026
MCP-Cosmos arXiv
MCP-Cosmos: World Model-Augmented Agents for Complex Task Execution in MCP Environments		 arXiv 2026
- arXiv
Generative Visual Code Mobile World Models		 arXiv 2026
IterCAD arXiv
IterCAD: An Iterative Multimodal Agent for Visually-Grounded CAD Generation and Editing		 arXiv 2026
- arXiv
Debugging Code World Models		 arXiv 2026
NeuralOS arXiv
NeuralOS: Towards Simulating Operating Systems via Neural Generative Models		 ICLR 2026
World-Model arXiv
World-Model-Augmented Web Agents with Action Correction		 arXiv 2026
WebWorld arXiv
WebWorld: A Large-Scale World Model for Web Agent Training		 arXiv 2026
WorldEvolver arXiv
Self-Evolving World Models for LLM Agent Planning		 arXiv 2026
AppDeltaWorld arXiv
AppDeltaWorld: Transition-Grounded Delta Code World Model for Mobile GUI Agents		 arXiv 2026
WebRider arXiv
WebRider: Persona-Conditioned Intent Controllers for Live-Web Assistance		 arXiv 2026
Twin arXiv
Twin: Playing an Unknown Game with a Test-Time Digital Twin		 arXiv 2026
WM-R1 arXiv
WM-R1: Training GUI Agents to Reason and leverage World Models with Reinforcement Learning		 arXiv 2026
- arXiv
Discriminative World Models for Web Agents		 arXiv 2026
EvoBelief arXiv
EvoBelief: Falsifiable World Models for Self-Improving LLM Agents		 arXiv 2026
CoworkEnv arXiv
CoworkEnv: Synthesizing Training Environments for Long-Horizon Cowork Agents Β· Code		 arXiv 2026
WildAWM arXiv
Agent World Model in the Wild: Computer-Use Agent Environment Synthesis Β· Code		 arXiv 2026

Memory, Experience & Skill Proxy

What prior experience, constraints, or reusable skills does the agent need now? Store-retrieve memory systems and reusable skill/behavior libraries.

Model Paper Venue
- arXiv
Dual-Layer Agentic Memory with Fast Write Routing and Slow Consolidation		 arXiv 2026
- arXiv
The Retriever Should Remember: Experience-Amortized Reranking for Long-Term Agent Memory		 arXiv 2026
- arXiv
Context as an Environment: Programmatic Context Management for Long-Horizon Agents		 arXiv 2026
- arXiv
ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL		 EMNLP 2026
- arXiv
MemFuse: Multi-Source Memory Fusion from Fragmented Observations		 arXiv 2026
- arXiv
Recursive Experiential-Working Memory Evolution for Long-Horizon Agent Harnesses		 arXiv 2026
- arXiv
TRACE: A Self-Evolving Skill Bank for Consistent, Limit-Aware LLM Agents		 arXiv 2026
- arXiv
CaSKG: Counterfactual-Causal Skill Graphs for Scalable Agent Skill Retrieval		 arXiv 2026
- arXiv
Coalition-Aware Skill Reliability for Self-Evolving Agents		 arXiv 2026
- arXiv
When Not to Imitate: Boundary-Aware Skill Memory for Reliable Tool-Use LLM Agents		 EMNLP 2026
- arXiv
Break It Down, Pass It On: Cross-Task Skill Transfer in LLM Agents		 arXiv 2026
- arXiv
WikiSkill: Compiling Agent Experience into Persistent Knowledge for Skill Evolution		 arXiv 2026
- arXiv
Auto-Policy, not Auto-Skill: Compiled Agent Skills for the Physical World		 arXiv 2026
- arXiv
SkillForge: Evolving Verifiable Skills for Reinforcement Learning Agents		 arXiv 2026
- arXiv
EVOMAL: Self-Poisoning in Self-Evolving Coding Agents		 arXiv 2026
- arXiv
Reasoning with Language Model is Planning with World Model		 EMNLP 2023
- arXiv
Generative Agents: Interactive Simulacra of Human Behavior		 UIST 2023
- arXiv
Cognitive Architectures for Language Agents		 TMLR 2024
Vismem arXiv
Vismem: Latent vision memory unlocks potential of vision-language models		 arXiv 2025
MemHarness arXiv
MemHarness: Memory Is Reconstructed, Not Replayed		 arXiv 2026
MemWM arXiv
MemWM: Memory-Augmented Text-Based World Model		 arXiv 2026
- arXiv
Activity Frames: Deterministic Screen-Activity Compilation for Agent Memory and Replay		 arXiv 2026
- arXiv
Governed Persistent Memory: Source-Bound State Semantics and Fail-Closed Release for Long-Horizon Agents		 arXiv 2026
PBD arXiv
PBD-AG: Persistent Baseline-Delta Active Graphs with Uncertainty-Aware Inspection for Long-Horizon Service Robots		 arXiv 2026
LUCID arXiv
LUCID: Latent-Skill Unified Control via Imagined Dynamics for Long-Horizon Humanoid Loco-Manipulation		 arXiv 2026
- arXiv
From Relevance to Execution Utility: Reward-Aware Dynamic Execution Gating for Skill-Based LLM Agents		 arXiv 2026
- arXiv
Progressive Experience Fusion for Multi-Task World Model Control in Endovascular Navigation		 arXiv 2026
MemoWM arXiv
MemoWM: How World Models Change What Agents Need to Remember Β· Code		 arXiv 2026
TallyMem arXiv
Beyond Retrieval: Evidence Access and Event Aggregation in Agent Memory		 arXiv 2026
MemComposer arXiv
MemComposer: Learning to Assemble Memories for LLM Agents via Reinforcement Learning		 arXiv 2026
AiO-Mem arXiv
All-in-One Memory: Process-Rewarded Reinforcement Learning for Memory Agents		 arXiv 2026
MemScope arXiv
MemScope: Evolving Agent Memory Architectures through Localized Code Repair		 arXiv 2026
TrackRL arXiv
Trackable Skill Library Lifecycle Management for Agentic Reinforcement Learning		 arXiv 2026

Reward / Verification Proxy

Is the agent's behavior correct, safe, and feasible, and how should it improve? Reward models, verifiers, critics, preference models, and LLM-as-judge.

Model Paper Venue
- arXiv
Grounded Checklist Partial Credit for Agent Skill Trajectories		 arXiv 2026
- arXiv
VICT: Verifier-Instrumented Credit Tracing for Long-Horizon LLM Agent Reinforcement Learning		 EMNLP 2026
- arXiv
Hindsight Memory-PRM: Supervising Memory Management with Auditable Hindsight Credit		 arXiv 2026
- arXiv
SRPO: Self-Reflective Policy Optimization for Long-Horizon Reasoning		 ICML 2026
- arXiv
Competence, Not Accuracy: A Diagnostic for Reference-Free Judge Gates in Skill Optimization		 arXiv 2026
- arXiv
Knowing When to Ask for Help: Bayesian Self-Escalation in Hierarchical LLM Agents		 arXiv 2026
- arXiv
When Stale Constraints Go Unchecked: Budgeted Verification Failures in Inherited Agent Memory		 arXiv 2026
- arXiv
MemGuard: Persisting Verifier Signals for LLM-Agent Memory Governance		 arXiv 2026
- arXiv
SPADE: Self-Play in Adaptive Synthetic Executable Environments		 arXiv 2026
- arXiv
EnvHarness: Awakening Static Worlds for Agent Learning		 arXiv 2026
- arXiv
LongWoF-Bench: Evaluating EvoMap Genes for Verifiable Long-Workflow Tasks		 arXiv 2026
- arXiv
Where World Models Break: Natural-Input Failure Discovery		 arXiv 2026
- arXiv
Deep Reinforcement Learning from Human Preferences		 NeurIPS 2017
Fine arXiv
Fine-Tuning Language Models from Human Preferences		 arXiv 2019
- arXiv
Training Verifiers to Solve Math Word Problems		 arXiv 2021
- arXiv
Direct Preference Optimization: Your Language Model is Secretly a Reward Model		 NeurIPS 2023
- arXiv
Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena		 NeurIPS 2023
- arXiv
A General Theoretical Paradigm to Understand Learning from Human Preferences		 AISTATS 2024
- arXiv
A Survey on LLM-as-a-Judge		 arXiv 2024
ORPO arXiv
ORPO: Monolithic Preference Optimization without Reference Model		 EMNLP 2024
- arXiv
Let's Verify Step by Step		 ICLR 2024
- arXiv
LLM Critics Help Catch LLM Bugs		 arXiv 2024
SimPO arXiv
SimPO: Simple Preference Optimization with a Reference-Free Reward		 NeurIPS 2024
WorldSimBench arXiv
WorldSimBench: Towards Video Generation Models as World Simulators		 arXiv 2024
WorldModelBench arXiv
WorldModelBench: Judging Video Generation Models As World Models		 arXiv 2025
Computer arXiv
Computer-Use Agents as Judges for Generative User Interface		 arXiv 2025
Critic-V arXiv
Critic-V: VLM Critics Help Catch VLM Errors in Multimodal Reasoning		 CVPR 2025
WCM arXiv
WCM: A World Critic Model for Vision-Language-Action Reinforcement Learning		 arXiv 2026
CheckVLA arXiv
CheckVLA: Execution-Time Verification with Action-Conditioned World Model for Long-Horizon Mobile Manipulation		 arXiv 2026
ContactGuard arXiv
ContactGuard: Pre-Contact Execution Monitoring with Action-Conditioned Latent World Models		 arXiv 2026
- arXiv
Failure Detection for Surgical Robot Imitation Policies via Flow-Matching World Modeling		 arXiv 2026
- arXiv
Calibrated Predictive Safety for Heterogeneous Robots: An Action-Conditioned JEPA Framework with Model-Based Safety Shields		 arXiv 2026
Ontology arXiv
Ontology-Grounded World Models for Failure Diagnosis and Closed-Loop Repair in Physical AI Systems		 arXiv 2026
StructReward arXiv
StructReward: Efficient Structured Process Rewards for Self-Correcting Multimodal Reasoning		 arXiv 2026
- arXiv
Why Study Emergent Behavior When You Can Regulate It? Aligning Multi-Agent Systems with Reward Prediction		 arXiv 2026
- arXiv
Refining Over Resampling: Test-Time Self-Correction for LLM Reasoning		 arXiv 2026
ARC arXiv
ARC: Fair Relative Advantage Comparison in Open-Ended Real-World Interaction		 arXiv 2026
LGWM arXiv
Right Screen, Wrong Transition: World Models as Verifiers for GUI Agents		 arXiv 2026
VeriVLA arXiv
VeriVLA: Adaptive Test-Time Verification for Flow-Based VLAs with Latent World Models		 arXiv 2026

Empowerment Levels (L1 / L2 / L3)

How proxies empower agents: L1 inference-time guidance (advise a frozen agent), L2 training-time optimization (generate experience/signal to train the agent), and L3 Agent-Proxy co-evolution (both improve each other in a loop).

Model Paper Venue
- arXiv
Scaling Agent Learning via Experience Synthesis		 arXiv 2025
RAGEN arXiv
RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learning		 arXiv 2025
VAGEN arXiv
VAGEN: Reinforcing World Model Reasoning for Multi-Turn VLM Agents		 NeurIPS 2025
WebEvolver arXiv
WebEvolver: Enhancing Web Agent Self-Improvement with Coevolving World Model		 EMNLP 2025
- arXiv
Aligning Agentic World Models via Knowledgeable Experience Learning		 arXiv 2026
- arXiv
Agent World Model: Infinity Synthetic Environments for Agentic Reinforcement Learning		 ICML 2026
SearchGym arXiv
SearchGym: Bootstrapping Real-World Search Agents via Cost-Effective and High-Fidelity Environment Simulation		 arXiv 2026
hint$^2$ arXiv
hint$^2$: Hierarchical World Models for Inference-Time Temporal Logic Guidance		 arXiv 2026

Evaluation, Fidelity & Safety

Benchmarks and metrics for world models, fidelity and calibration limits, and the safety/attack surface of proxy-driven learning.

Model Paper Venue
G-Eval arXiv
G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment		 EMNLP 2023
Length arXiv
Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators		 COLM 2024
Prometheus arXiv
Prometheus: Inducing Fine-Grained Evaluation Capability in Language Models		 ICLR 2024
BEHAVIOR-1K arXiv
BEHAVIOR-1K: A Human-Centered, Embodied AI Benchmark with 1,000 Everyday Activities and Realistic Simulation		 arXiv 2024
- arXiv
Evaluating the World Model Implicit in a Generative Model		 NeurIPS 2024
- arXiv
World Models: The Safety Perspective		 arXiv 2024
- arXiv
Understanding World or Predicting Future? A Comprehensive Survey of World Models		 CSUR 2025
- arXiv
How Far is Video Generation from World Model: A Physical Law Perspective		 ICML 2025
- arXiv
When World Models Dream Wrong: Physical-Conditioned Adversarial Attacks against World Models		 arXiv 2026
- arXiv
Safety in Embodied AI: A Survey of Risks, Attacks, and Defenses		 arXiv 2026
JailWAM arXiv
JailWAM: Jailbreaking World Action Models in Robot Control		 arXiv 2026
- arXiv
Current Agents Fail to Leverage World Model as Tool for Foresight		 arXiv 2026
CtrlAttack arXiv
CtrlAttack: A Unified Attack on World-Model Control in Diffusion Models		 arXiv 2026
SafeDream arXiv
SafeDream: Safety World Model for Proactive Early Jailbreak Detection		 arXiv 2026
KineBench arXiv
KineBench: Benchmarking Embodied World Models via IDM-Free Kinematic Grounding		 arXiv 2026
Apple arXiv
Apple-$Ο€$: Benchmarking Thinking with Video Towards Law-Grounded Physical Intelligence		 arXiv 2026
WorldExam arXiv
WorldExam: Benchmarking World Models from Apparent Appearance to Inherent Reactivity		 arXiv 2026
GAUGE arXiv
GAUGE: A Measurement-Grounded Benchmark for Physical Fidelity in Simulation Engines and Video World Models		 arXiv 2026
H2R arXiv
H2R-Bench: Benchmarking Human-to-Robot Manipulation Video Generation in World Models		 arXiv 2026
WorldSimProbe arXiv
WorldSimProbe: Diagnosing Simulator Faithfulness in Action-Conditioned World Models for Embodied Manipulation		 arXiv 2026
VIScore arXiv
VIScore: Diagnosing Planning-Relevant Quality in Latent World Models		 arXiv 2026
HarnessEval arXiv
HarnessEval-W: Agentifying the Evaluation of Visual Worlds		 arXiv 2026
CG arXiv
CG-World: A Large-Scale World-State Dataset and Protocol for World Models		 arXiv 2026
- arXiv
What Can Latent World Models Know? Physical Parameter Identifiability in Multimodal Predictive Representations		 arXiv 2026
- arXiv
Ensuring Safe Physical AI in Urban Mobility via Hazard-Informed Synthesized Envelopes		 arXiv 2026
StateAliasBench arXiv
The GUI Is Not the State: Diagnosing State Aliasing in GUI World Models		 arXiv 2026
GlobeBench arXiv
GlobeBench: Evaluating Text-Based World Model Capabilities for Scalable Environment Simulations Β· Code 		arXiv 2026
AnyGoal arXiv
AnyGoal: On World Model Generalization Beyond Expert Rollouts Β· Code Β· Project 		arXiv 2026

Surveys & Related Work

Related surveys and the paper's core position statements.

Model Paper Venue
- arXiv
A Systematic Survey of Agentic Skills: Architecture, Lifecycle, and Security		 arXiv 2026
- arXiv
Is Sora a World Simulator? A Comprehensive Survey on General World Models and Beyond		 arXiv 2024
- arXiv
A Survey of World Models for Autonomous Driving		 arXiv 2025
- arXiv
A Comprehensive Survey on World Models for Embodied AI		 arXiv 2025
- arXiv
Agentic World Modeling: Foundations, Capabilities, Laws, and Beyond		 arXiv 2026
- arXiv
Reinforcement Learning: From Algorithms To Foundation Models		 arXiv 2026
Planning arXiv
Planning-Oriented End-to-End Autonomous Driving: Architectures, Evaluation, and Emerging Paradigms		 arXiv 2026
- arXiv
Toward the Cognitive--Physical Limits of Embodied Intelligence through a World-Model-Centric Autonomous Racing Agent		 arXiv 2026

Additional References

Further supporting works cited in the survey.

Model Paper Venue
- arXiv
Proximal Policy Optimization Algorithms		 arXiv 2017
Habitat Habitat: A Platform for Embodied AI Research ICCV 2019
- Model-Based Reinforcement Learning for Atari ICLR 2020
- arXiv
Learning to Summarize from Human Feedback		 NeurIPS 2020
- arXiv
Transporter Networks: Rearranging the Visual World for Robotic Manipulation		 CoRL 2020
CLIPort arXiv
CLIPort: What and Where Pathways for Robotic Manipulation		 CoRL 2021
- arXiv
Do As I Can, Not As I Say: Grounding Language in Robotic Affordances		 CoRL 2022
- arXiv
Constitutional AI: Harmlessness from AI Feedback		 arXiv 2022
- arXiv
Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback		 arXiv 2022
TensoRF arXiv
TensoRF: Tensorial Radiance Fields		 ECCV 2022
- arXiv
Inner Monologue: Embodied Reasoning through Planning with Language Models		 CoRL 2022
- arXiv
Large Language Models are Zero-Shot Reasoners		 NeurIPS 2022
- arXiv
Instant Neural Graphics Primitives with a Multiresolution Hash Encoding		 TOG 2022
- arXiv
Training Language Models to Follow Instructions with Human Feedback		 NeurIPS 2022
Perceiver-Actor arXiv
Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulation		 CoRL 2022
- arXiv
Solving Math Word Problems with Process- and Outcome-Based Feedback		 arXiv 2022
Chain-of arXiv
Chain-of-Thought Prompting Elicits Reasoning in Large Language Models		 NeurIPS 2022
Plenoxels arXiv
Plenoxels: Radiance Fields without Neural Networks		 CVPR 2022
Self arXiv
Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture		 CVPR 2023
RT-1 arXiv
RT-1: Robotics Transformer for Real-World Control at Scale		 RSS 2023
Self-Refine arXiv
Self-Refine: Iterative Refinement with Self-Feedback		 NeurIPS 2023
MemGPT arXiv
MemGPT: Towards LLMs as Operating Systems		 arXiv 2023
Reflexion arXiv
Reflexion: Language Agents with Verbal Reinforcement Learning		 NeurIPS 2023
ReAct arXiv
ReAct: Synergizing Reasoning and Acting in Language Models		 ICLR 2023
- arXiv
Tree of Thoughts: Deliberate Problem Solving with Large Language Models		 NeurIPS 2023
RRHF arXiv
RRHF: Rank Responses to Align Language Models with Human Feedback without Tears		 NeurIPS 2023
KTO arXiv
KTO: Model Alignment as Prospect Theoretic Optimization		 ICML 2024
SWE-bench arXiv
SWE-bench: Can Language Models Resolve Real-World GitHub Issues?		 ICLR 2024
Oasis Oasis: A Universe in a Transformer 2024
- arXiv
Agent Planning with World Knowledge Model		 NeurIPS 2024
DeepSeekMath arXiv
DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models		 arXiv 2024
VGGT arXiv
VGGT: Visual Geometry Grounded Transformer		 arXiv 2025
MonST3R arXiv
MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motion		 ICLR 2025
- arXiv
The latent space: Foundation, evolution, mechanism, ability, and outlook		 arXiv 2026

Acknowledgements

This list is maintained alongside the survey. Thanks to all contributors and to the authors of the works listed here.

Citation

If you find this work useful, please consider citing:

@article{yang2026worldproxy,
  title   = {Quo Vadis, World Modeling?},
  author  = {Yu Yang and Xuemeng Yang and Licheng Wen and Lingdong Kong and Xiaobin Hu and Dongyue Lu and Wei Chow and Xiyan Huang and Yuxiang Feng and Yue Liao and Jianbiao Mei and Daocheng Fu and Rong Wu and Pinlong Cai and Ran Yi and Ying Tai and Jiangning Zhang and Botian Shi and Yong Liu and Shuicheng Yan},
  journal = {arXiv preprint arXiv:2608.02713},
  year    = {2026}
}

About

πŸ”₯ Quo Vadis, World Modeling? Towards Interactive World Proxies for Continually Improving Agents

worldbench.github.io/awesome-agentic-world-model

Topics

agentagenticagentic-aiagentic-world-modelagentsaigcclosed-loopclosed-loop-controlclosed-loop-experimentsclosed-loop-simulationgenerationgenerative-aiimage-generationvideo-generationworld-modelworld-modelingworld-modelsworld-models-rl

Resources

Readme
MIT license
Activity
Custom properties

Stars

62 stars

Watchers

2 watching

Forks

1 fork
Report repository

Releases

Packages

Contributors

Languages