Spatial AI
World Labs LSM and Atlas
World Labs has moved the public conversation from a generic LSM label toward Atlas, an omni world model for spatial intelligence announced on September 1, 2026.
Use this workspace to turn a loose idea into a focused video brief with a subject, motion direction, reference style, and next scene cue.
World Labs' latest official update names Atlas as a world model for spatial intelligence.
The long-term topic is spatial intelligence: generating and reasoning about explorable worlds, not only flat video.
This page keeps the LSM keyword while using the fresher Atlas naming where it matters.
What a Large Spatial Model means
A Large Spatial Model is best understood as an AI system that represents scenes as spaces people or agents can move through, inspect, and reason about.
Why Atlas changes the page
Atlas gives this topic a current product name. For visitors, the useful question is what spatial AI can do that a normal video generator cannot.
Spatial AI vs AI video
Seedance, Kling, and H3 Max focus on video output. Spatial AI points toward persistent worlds where camera control and geometry are part of the result.
FAQ
Is World Labs LSM a tool I can use today?
Public access changes over time. This page is a guide and tracker, with links to World Labs' official updates for the current availability status.
Why mention Atlas on an LSM page?
Because Atlas is World Labs' current public naming for its world-model direction, while many searchers still use the older Large Spatial Model phrase.