Short extractable summary preserved directly in static HTML.
Editorial provenance and refresh policy preserved directly in static HTML.
Published by world-models.io editorial board.
Lead editor Bernard Grenat.
This hub maintains a structured directory of world model records connected to labs, categories, comparisons, and research pages.
Each editorial page is assembled from primary sources, normalized into extractable summaries, checked for factual drift, and reviewed before publication or major refreshes. Last reviewed: 2026-06-21.
Pages are refreshed when a new paper, benchmark, release, architecture update, or stronger primary source materially changes the answer a reader or AI system should retrieve.
Each page links back to relevant primary sources and keeps a stable canonical URL so readers can verify claims, trace context, and reference the most up-to-date version. See the editorial policy.
Canonical static snapshot of the world models database. Client-side filters can still enhance this page for interactive users.
| Model | Lab | Category | Status | Modality | Year |
|---|---|---|---|---|---|
| HY-World 2.0 | Tencent Hunyuan | Foundation World Model | active | Text/Image/Video → Navigable 3D World | 2026 |
| LeWorldModel | Mila / NYU / Samsung SAIL | Self-Supervised World Model | active | Visual → Latent Predictions | 2026 |
| Marble | World Labs | Foundation World Model | active | Text/Image/Video/360 -> Persistent 3D World | 2026 |
| PixVerse R1 | PixVerse | Generative World Model | active | Text/Image → Interactive Shared World | 2026 |
| PlayWorld | Princeton University | Generative World Model | active | Multi-view Robot Video + Actions -> Future Observations | 2026 |
| 1X World Model | 1X | Foundation World Model | active | Robot Video + Actions -> Future Observations | 2025 |
| GAIA-2 | Wayve | Generative World Model | active | Multi-View Driving Video + Structured Conditions → Future Driving Video | 2025 |
| Genie 3 | Google DeepMind | Generative World Model | active | Text → Interactive 3D Environment | 2025 |
| Matrix-Game 2.0 | Skywork AI | Generative World Model | active | Image + Keyboard/Mouse Actions → Streaming Interactive Video | 2025 |
| Odyssey-2 | Odyssey | Foundation World Model | active | Text + Video Context → Interactive Video Stream | 2025 |
| RELIC | Adobe Research | Generative World Model | emerging | Image + Text + Camera Actions → Interactive Video | 2025 |
| V-JEPA 2 | Meta | Self-Supervised World Model | active | Video → Latent Predictions | 2025 |
| WHAM | Microsoft Research / Ninja Theory | Generative World Model | active | Game Video + Controller Actions → Future Video + Actions | 2025 |
| WHAM-RT | Microsoft Research / Ninja Theory | Generative World Model | active | Game State + Actions → Real-Time Interactive Video | 2025 |
| 3D-VLA | MIT / Tsinghua | Foundation World Model | emerging | 3D Point Clouds + Language + Actions | 2024 |
| AMI World Model | AMI Labs | Foundation World Model | emerging | Vision + Language + Proprioception | 2024 |
| DIAMOND | Microsoft Research / University of Geneva | Model-Based RL | active | Visual | 2024 |
| GameNGen | Google Research | Generative World Model | active | Actions + Frame History → Next Frame | 2024 |
| Gen-3 Alpha | Runway | Generative World Model | active | Text + Image → Video | 2024 |
| Genie | Google DeepMind | Generative World Model | foundational | Image → Interactive 2D Environment | 2024 |
| Genie 2 | Google DeepMind | Generative World Model | active | Image → Interactive 3D Environment | 2024 |
| Large World Model (LWM) | UC Berkeley | Foundation World Model | active | Visual (Video) + Language | 2024 |
| NVIDIA Cosmos | NVIDIA | Foundation World Model | active | Video + 3D | 2024 |
| OASIS | Decart / Etched | Generative World Model | active | Actions → Video (real-time) | 2024 |
| Pandora | Tsinghua University / ByteDance | Generative World Model | emerging | Text + Actions → Video | 2024 |
| Sora | OpenAI | Generative World Model | active | Text → Video | 2024 |
| TD-MPC2 | MIT / Meta | Model-Based RL | active | State + Visual | 2024 |
| V-JEPA | Meta | Self-Supervised World Model | active | Video | 2024 |
| Copilot4D | Waabi | Foundation World Model | active | LiDAR point clouds (4D) | 2023 |
| DreamerV3 | Google DeepMind | Model-Based RL | active | Visual + Proprioceptive | 2023 |
| Emu Video | Meta | Generative World Model | active | Text → Image → Video | 2023 |
| GAIA-1 | Wayve | Foundation World Model | active | Video + Text + Actions | 2023 |
| I-JEPA | Meta | Self-Supervised World Model | active | Image | 2023 |
| IRIS | Microsoft Research | Model-Based RL | active | Visual | 2023 |
| RT-2 | Google DeepMind | Foundation World Model | active | Visual + Language + Proprioceptive | 2023 |
| Stable Video Diffusion | Stability AI | Generative World Model | active | Visual (Image → Video) | 2023 |
| STEVE-1 | UT Austin | Generative World Model | active | Visual + Language (instructions) | 2023 |
| UniSim | Google DeepMind | Generative World Model | active | Video + Actions | 2023 |
| MILE | Wayve | Foundation World Model | active | Visual (Multi-camera) + Ego-state | 2022 |
| Waabi World | Waabi | Generative World Model | active | Driving Data + Agent Actions → Reactive Driving Simulation | 2022 |
| DreamerV2 | Model-Based RL | foundational | Visual | 2021 | |
| MuZero | Google DeepMind | Model-Based RL | foundational | Board states / Visual (Atari) | 2020 |
| PlaNet | Model-Based RL | foundational | Visual | 2019 | |
| RSSM | Latent Dynamics | active | Visual + Proprioceptive | 2019 | |
| World Models (Ha & Schmidhuber) | Google Brain / IDSIA | Model-Based RL | foundational | Visual | 2018 |
| Imagination-Augmented Agents (I2A) | Google DeepMind | Model-Based RL | foundational | Visual | 2017 |
| Predictron | Google DeepMind | Model-Based RL | foundational | Abstract state representations | 2017 |
| Value Prediction Network (VPN) | University of Michigan / Google Brain | Model-Based RL | foundational | Visual / Abstract states | 2017 |
Representative frontier systems highlighted directly in static HTML.
| Model | Lab | Category | Summary |
|---|---|---|---|
| HY-World 2.0 | Tencent Hunyuan | Foundation World Model | Tencent Hunyuan's open multimodal world model for reconstructing, generating, and simulating navigable 3D worlds. |
| LeWorldModel | Mila / NYU / Samsung SAIL | Self-Supervised World Model | A compact 15M-parameter JEPA world model that learns real-world physics on a single GPU, solving the notorious representation collapse problem. |
| Marble | World Labs | Foundation World Model | World Labs' multimodal world model for generating spatially consistent, persistent 3D environments from text, images, video, and 360 inputs. |
| PixVerse R1 | PixVerse | Generative World Model | The first real-time world model supporting multi-user shared worlds with personalized avatars and no session limits. |
| PlayWorld | Princeton University | Generative World Model | A Princeton robot world model trained from autonomous self-play to simulate contact-rich manipulation and support policy evaluation and RL fine-tuning. |
| 1X World Model | 1X | Foundation World Model | 1X's physics-grounded video world model for anticipating the outcomes of NEO's actions and supporting generalization to unseen household tasks. |
Static usage guidance for non-JS readers.
Use category and lab pages to narrow the field, then open individual model pages for architecture details, benchmarks, citations, and related entities.
For ranked browsing, continue to the leaderboard and the Performance Index methodology. For editorial curation, use the top-world-models and robotics collections.
FAQ answers rendered directly into static HTML for extractable responses.
The static database currently includes 48 world model records sourced from the local knowledge base.
You can compare model category, originating lab, lifecycle status, modality, and publication year before opening detail pages or side-by-side comparisons.
Short extractable summary preserved directly in static HTML.
Editorial provenance and refresh policy preserved directly in static HTML.
Published by world-models.io editorial board.
Lead editor Bernard Grenat.
This database maintains structured model records connected to labs, categories, comparisons, and research pages for retrieval and verification.
Each editorial page is assembled from primary sources, normalized into extractable summaries, checked for factual drift, and reviewed before publication or major refreshes. Last reviewed: 2026-06-21.
Pages are refreshed when a new paper, benchmark, release, architecture update, or stronger primary source materially changes the answer a reader or AI system should retrieve.
Each page links back to relevant primary sources and keeps a stable canonical URL so readers can verify claims, trace context, and reference the most up-to-date version. See the editorial policy.
Primary references and official sources surfaced directly in static HTML for crawlers and no-JS readers.