Meta's RoboJEPA: 8B-Param Robot World Model With Scaling Laws
arankomatsuzaki · x · 2026-10-09
RoboJEPA: Scaling Robotic Latent World Models
Meta introduces RoboJEPA, a world model built on the Joint Embedding Predictive Architecture (JEPA), scaled from 22M to 8B parameters — the largest JEPA predictor trained to date — using data spanning 12 robotic embodiments.
Key findings:
- The model's "imagination error" (error of latent rollouts) follows a second-order power law in compute, enabling quality predictions well beyond the scale where the law is fit.
- Downstream robotic planning performance improves predictably with compute and correlates strongly with imagination error, making it a reliable proxy for expensive real-robot evaluation.
- Latent world models can be deployed zero-shot as robotic agents, planning toward a single goal image to solve long-horizon tasks on real hardware.
- This is the first work to establish scaling laws for multi-embodiment robot world models trained on real robot data.
The paper (arXiv:2610.10515) releases all model checkpoints plus training and robot deployment code.
Related event: Meta Releases RoboJEPA, an 8B-Parameter Robotic World Model(3 posts)→
More from Embodied
- Rider shares Cybercab experience: watching YouTube, 'doesn't want the ride to end' — elonmusk · 2026-10-09
- RoboStrategy to Ring Nasdaq Closing Bell with Dyna Robotics Humanoid at Founders Forum — Rewkang · 2026-10-09
- New Paper: Native Action-Prior Learning from Videos for World Action Models — _akhaliq · 2026-10-09
- Shenzhen Hotel Robot Delivers Surprise Birthday Cake, But "No Human Involved" Isn't Quite True — EleanorOlcott · 2026-10-09
- Open-Source Models Power Home Robot Tidying for Toddlers in Weeks, Not Years — chris_j_paxton · 2026-10-09
- Sony's quadruped walks on cracked pavement with passive-stability spherical feet — BLUECOW009 · 2026-10-09