← Back to Videos

SmolWAM: A World-Action Model Shrunk to 57ms on the Edge

Loading video

SmolWAM shrinks a world-action model to 57ms edge inference while still forecasting future states, benchmarked side-by-side against Cosmos Policy 2B on LIBERO tasks.

World Models Edge computing VLA LIBERO

Category: world-model

Author: @robotsdigest

Date: 2026-10-13T00:00:00

Duration: 67.011s

Source: https://x.com/robotsdigest/status/2110051826095808934

Multics: Multi-Player World Model Scaling to 10 Players

Multics: Multi-Player World Model Scaling to 10 Players

@XihuiLiu · 87 views · 2026-10-14

World ModelsVideo GenerationMulti-Player

Odyssey 3: World Models Let Robots Rehearse Physics

Odyssey 3: World Models Let Robots Rehearse Physics

@joaquintenest · 237 views · 2026-10-09

World ModelsSim-to-RealRobot training

XPENG XPACE: A Unified Embodied World Model for IRON

XPENG XPACE: A Unified Embodied World Model for IRON

@CyberRobooo · 667 views · 2026-09-17

World ModelsEmbodied AIXpeng

PointZero: 3D World Model Pretrained Without Robot Data

PointZero: 3D World Model Pretrained Without Robot Data

@bardienus · 566 views · 2026-09-17

World Models3D DynamicsPoint Tracking

OpenWAM open world-action model pretraining framework

OpenWAM open world-action model pretraining framework

@KnightNemo_ · 588 views · 2026-09-09

World ModelsRobot LearningPretraining

AgiBot GE-Act 2.0: A World Action Model Trained From Scratch

AgiBot GE-Act 2.0: A World Action Model Trained From Scratch

@Chinazhidx · 671 views · 2026-09-09

World-action modelsVLARobot Manipulation

As an Amazon Associate, we earn from qualifying purchases.