Main comparison summary preserved directly in static HTML.
The original 2018 'World Models' paper vs. the current state-of-the-art: how five years of research transformed a foundational concept into a domain-general world model agent.
Primary editorial conclusion preserved for non-JS crawlers and readers.
Ha & Schmidhuber's World Model is where it all started: the paper that popularized the idea of training in imagination. DreamerV3 is where it is now: a mature system that masters diverse domains with fixed hyperparameters. Studying both provides the clearest picture of how the field evolved from elegant concept to practical mastery.
Extractable difference list generated from the comparison table.
Static decision guidance for no-JS readers.
Choose Ha & Schmidhuber World Model when its capabilities best match your research or deployment requirements.
Choose DreamerV3 when its capabilities best match your research or deployment requirements.
Ha & Schmidhuber's World Model is where it all started: the paper that popularized the idea of training in imagination. DreamerV3 is where it is now: a mature system that masters diverse domains with fixed hyperparameters. Studying both provides the clearest picture of how the field evolved from elegant concept to practical mastery.
| Dimension | Ha & Schmidhuber World Model | DreamerV3 |
|---|---|---|
| Architecture | VAE + MDN-RNN + Controller | RSSM with actor-critic |
| Latent Space | Continuous VAE | Discrete categorical |
| Domains | Car Racing, VizDoom | Atari, DMControl, Minecraft, etc. |
| Policy | Linear controller (CMA-ES) | Neural actor-critic |
| Sample Efficiency | Moderate | Very high |
| Historical Role | Seminal: coined 'World Models' | Current SOTA for general model-based RL |
| Year | 2018 | 2023 |
High-level scoring context for the models referenced in this comparison.
| Model | Category | Index v1.1 | Confidence |
|---|---|---|---|
| World Models (Ha & Schmidhuber) | Model-Based RL | 48/100 | high |
| DreamerV3 | Model-Based RL | 88/100 | high |
| PlaNet | Model-Based RL | 57/100 | high |
| DreamerV2 | Model-Based RL | 72/100 | high |
FAQ answers rendered directly into static HTML for extractable responses.
Absolutely. Ha & Schmidhuber's paper is simpler and more conceptually clear. It provides the best introduction to world model concepts before studying the more complex Dreamer family.
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 comparison page publishes a direct answer, explicit trade-offs, and source-backed evidence that can be validated against primary materials.
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 papers and official sources for the models discussed on this comparison page.