---
title: "LeWorldModel | Compact JEPA World Model | world-models.io"
canonical: "https://world-models.io/en/models/leworldmodel/"
description: "LeWorldModel is a compact 15M-parameter JEPA world model that learns physics on a single GPU, solving the representation collapse problem."
author: "Tyler D."
generated: "2026-10-30"
---

New: the Timeline is live. Track world model releases, papers, and benchmark updates in real time.

The Knowledge Hub for AI World Models

# LeWorldModel

LeWorldModel (LeWM) is a compact 15M-parameter JEPA architecture that learns physical reality on a single GPU, planning up to 48x faster than foundation models. Introduces SigReg to solve the JEPA collapse problem.

robotics

model-based-rl

simulation

embodied-ai

## What is a world model?

A world model is an AI system that learns a predictive representation of how an environment changes over time. It can use observations, actions, or language to estimate future states and possible outcomes before an agent acts. World models support planning in model-based reinforcement learning, robotics, embodied AI, autonomous driving, and interactive simulation. This site catalogs these systems with model records, research papers, benchmark context, and source links. Start with the [world models database](https://world-models.io/en/world-models-database/), compare systems on the [leaderboard](https://world-models.io/en/leaderboard/), or read the [Performance Index methodology](https://world-models.io/en/methodology/performance-index/).

## Key Attributes

Static HTML snapshot of the model record for crawlers and no-JS readers.

| Attribute | Value |
| --- | --- |
| Model | LeWorldModel |
| Lab / Organization | Mila / NYU / Samsung SAIL |
| Category | Self-Supervised World Model |
| Subtype | Compact JEPA Model |
| World Model Type | Compact joint-embedding predictive world model for physical understanding |
| Primary Domain | Physical Reasoning |
| Architecture | Compact Vision Transformer with JEPA and SigReg regularization |
| Modality | Visual → Latent Predictions |
| Training Method | Self-supervised JEPA training with SigReg collapse prevention, single-GPU trainable |
| Status | active |
| Year | 2026 |
| Performance Index | 78/100 (medium confidence, v1.1) |

## About LeWorldModel

Main editorial body preserved directly in static HTML.

LeWorldModel (LeWM) is a compact JEPA-based world model with only 15 million parameters that achieves physical understanding and planning up to 48× faster than foundation-model-based alternatives. Developed by researchers from Mila, NYU, and Samsung SAIL, it introduces SigReg, a novel regularization technique that solves the representation collapse problem that has plagued JEPA architectures. LeWM demonstrates that large-scale foundation models are not strictly necessary for physical reasoning: efficient, well-regularized architectures can learn meaningful world representations from pixels on commodity hardware.

LeWorldModel is a compact joint-embedding predictive world model for physical understanding developed by Mila / NYU / Samsung SAIL in 2026 for physical reasoning.

## Editorial Snapshot

Short extractable facts for answer engines and no-JS readers.

| Signal | Value |
| --- | --- |
| Definition | LeWorldModel is a compact joint-embedding predictive world model for physical understanding developed by Mila / NYU / Samsung SAIL in 2026 for physical reasoning. |
| Short Description | A compact 15M-parameter JEPA world model that learns real-world physics on a single GPU, solving the notorious representation collapse problem. |
| Benchmark Rows | 2 |
| FAQ Entries | 1 |
| Related Models | 4 |
| Related Guides | 0 |
| Related Research Topics | 0 |
| Last Updated | 2026-04-10 |

## Notable Features

Key capabilities associated with this model.

- • Only 15M parameters
- • Trainable on a single GPU
- • SigReg solves JEPA collapse problem
- • 48× faster planning than foundation models
- • End-to-end from pixels

## Use Cases

Representative applications attached to this model record.

Physical scene understanding

Efficient robot planning

Resource-constrained deployment

Research prototyping

## Strengths and Limitations

Balanced assessment surfaced in static HTML.

### Strengths

- • Extremely compact
- • Single-GPU training
- • Solves collapse problem
- • Fast inference
- • Open research

### Limitations

- • Limited to simple physical scenarios
- • Early-stage research
- • Not yet tested at scale

## Benchmarks

Published benchmark evidence attached to this model record.

| Benchmark | Metric | Result | Source |
| --- | --- | --- | --- |
| Planning Efficiency | Planning Speedup 48 x | Up to 48x faster | [Source](https://arxiv.org/abs/2603.19312) |
| Training Footprint | Hardware Requirement 1 GPU | Single-GPU trainable | [Source](https://arxiv.org/abs/2603.19312) |

## References and Citations

Primary references preserved in static HTML for citation extraction.

| Reference | Link |
| --- | --- |
| LeCun et al., 2026. LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels. arXiv:2603.19312. | [Open source](https://arxiv.org/abs/2603.19312) |

## Related Models

Nearby models linked from the current editorial record.

| Model | Category | World Model Type | Index v1.1 |
| --- | --- | --- | --- |
| [V-JEPA](https://world-models.io/en/models/v-jepa/) | Self-Supervised World Model | Self-supervised visual world model | 70/100 |
| [V-JEPA 2](https://world-models.io/en/models/v-jepa-2/) | Self-Supervised World Model | Joint-embedding predictive world model for video understanding and robot planning | 87/100 |
| [I-JEPA](https://world-models.io/en/models/i-jepa/) | Self-Supervised World Model | Self-supervised visual world model | 61/100 |
| [AMI World Model](https://world-models.io/en/models/ami-world-model/) | Foundation World Model | Announced multimodal world model initiative | N/A |

## Direct Comparisons

Side-by-side comparisons already connected to this model.

| Comparison | Matchup | Summary |
| --- | --- | --- |
| [V-JEPA 2 vs V-JEPA](https://world-models.io/en/compare/v-jepa-2-vs-v-jepa/) | V-JEPA 2 vs V-JEPA | V-JEPA 2 dramatically scales up Meta FAIR's self-supervised video world model, achieving state-of-the-art visual understanding and zero-shot robot control, capabilities V-JEPA didn't demonstrate. |
| [V-JEPA 2 vs I-JEPA](https://world-models.io/en/compare/v-jepa-2-vs-i-jepa/) | V-JEPA 2 vs I-JEPA | Two milestones of the JEPA roadmap: I-JEPA established self-supervised image representation by predicting in latent space; V-JEPA 2 extends the paradigm to video at foundation scale and demonstrates zero-shot robot control. |
| [LeWorldModel vs DreamerV3](https://world-models.io/en/compare/leworldmodel-vs-dreamer-v3/) | LeWorldModel vs DreamerV3 | LeWorldModel revisits LeCun's energy-based JEPA philosophy for control, predicting in latent space without pixel reconstruction. DreamerV3 remains the canonical RSSM-based agent that learns by imagining pixel-grounded rollouts. |
| [Genie 3 vs V-JEPA 2](https://world-models.io/en/compare/genie-3-vs-v-jepa-2/) | Genie 3 vs V-JEPA 2 | Two green-index leaders that represent different frontier philosophies. Genie 3 is an interactive generative world model that turns text into playable environments, while V-JEPA 2 is a self-supervised latent predictor optimized for physical reasoning and zero-shot robot planning. |

## Frequently Asked Questions

FAQ answers rendered directly into static HTML for extractable responses.

### What is SigReg?

SigReg is a regularization technique introduced in LeWorldModel that prevents the representation collapse problem in JEPA architectures by constraining the singular values of the latent representations, ensuring they remain informative throughout training.

## Quick Answer

Short extractable summary preserved directly in static HTML.

- LeWorldModel is a compact joint-embedding predictive world model for physical understanding developed by Mila / NYU / Samsung SAIL in 2026 for physical reasoning.
- Use this page when you need a fast read on how LeWorldModel fits into the self-supervised world model landscape, then validate the details in the benchmarks, citations, and related pages.
- A key strength surfaced in the editorial record is extremely compact.

## Editorial Trust Signals

Editorial provenance and refresh policy preserved directly in static HTML.

**Published by** world-models.io editorial board.

**Lead editor** Tyler D. - Technical editor, methodology and benchmark analysis.

This model page synthesizes primary papers, official model pages, benchmark evidence, and related world-models.io context into a reference resource.

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: 2026-04-10.

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.](https://world-models.io/en/editorial-policy/)

Primary sources only

Last reviewed date visible

Methodology documented

Source links included

Correction channel published

Review date visible

## External Sources

Primary model and lab sources embedded in static HTML.

- [LeWorldModel paper](https://arxiv.org/abs/2603.19312) Mila / NYU / Samsung SAIL primary research source

## References

1. [[1] LeCun et al., 2026. LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels. arXiv:2603.19312.](https://arxiv.org/abs/2603.19312)