← World Models

Simulation & Driving

Foundation world models, interactive video generation, game simulation, and autonomous driving world models.

⌘K

World models aim to learn internal representations of how the world works, enabling agents to predict future states, plan actions, and reason about counterfactuals without direct interaction.

Foundation World Models 10 models

ModelFull TitleRepresentationYear
Cosmos 2.5Cosmos World Foundation Models: Scaling Physical AI Video Generation2025
Genie 3Genie 3: Interactive World Models at Scale Interactive Video2025
V-JEPA 2V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction, and Planning Latent (JEPA)2025
PANA World Model for General, Interactable, and Long-Horizon World Simulation Interactive2025
SoraVideo generation models as world simulators Video Generation2024
Genie 2A Large-Scale Foundation World Model Interactive Video2024
GenieGenerative Interactive Environments Interactive Video2024
CosmosCosmos World Foundation Model Platform for Physical AI Video Generation2024
UniSimLearning Interactive Real-World Simulators Interactive2023
PandoraTowards General World Model with Natural Language Actions and Video States Interactive2024

Game Simulation 11 models

ModelFull TitleSpace
GameNGenDiffusion Models Are Real-Time Game Engines Pixel
DIAMONDDiffusion for World Modeling: Visual Details Matter in Atari Pixel
Matrix-GameInteractive World Foundation Model Pixel
Matrix-Game 2.0Open-Source, Real-Time, and Streaming Interactive World Model Pixel
OasisA Universe in a Transformer Pixel
MineWorldA Real-Time and Open-Source Interactive World Model on Minecraft Pixel
AnimeGamerInfinite Anime Life Simulation with Next Game State Prediction Pixel
GameFactoryCreating New Games with Generative Interactive Videos Pixel
WORLDMEMLong-term Consistent World Simulation with Memory Pixel
HunyuanWorld 1.0Generating Immersive, Explorable, and Interactive 3D Worlds 3D Mesh
Matrix-3DOmnidirectional Explorable 3D World Generation 3D Mesh

Autonomous Driving 16 models

ModelFull Title
ResWorldResWorld: Temporal Residual World Model for End-to-End Autonomous Driving
DrivingGenDrivingGen: Comprehensive Benchmark for Generative Video World Models in Autonomous Driving
Cosmos-Drive-DreamsScalable Synthetic Driving Data Generation with World Foundation Models
GAIA-2A Controllable Multi-View Generative World Model for Autonomous Driving
GAIA-1Introducing a Cutting-Edge Generative AI Model for Autonomy
VistaA Generalizable Driving World Model with High Fidelity and Versatile Controllability
DrivingWorldConstructing World Model for Autonomous Driving via Video GPT
OccWorldLearning a 3D Occupancy World Model for Autonomous Driving
Drive-WMDriving into the Future: Multiview Visual Forecasting and Planning with World Model
CarDreamerOpen-Source Learning Platform for World Model based Autonomous Driving
SafeDreamerSafe Reinforcement Learning with World Models
GenADGeneralized Predictive Model for Autonomous Driving
Copilot4DLearning Unsupervised World Models for Autonomous Driving via Discrete Diffusion
DrivingGPTUnifying Driving World Modeling and Planning with Multi-modal Autoregressive Transformers
Doe-1Closed-Loop Autonomous Driving with Large World Model
GEMA Generalizable Ego-Vision Multimodal World Model for Fine-Grained Control