Viggle Launches JST-2, a Physics-Native 3D World Foundation Model

Viggle released its new model JST-2 on October 20, billing it as the world's first full-stack "physics-native" world foundation model. Unlike most approaches that learn world models in pixel/video space, JST-2 starts from first principles and builds the entire model stack directly on 3D — a route that sets it clearly apart from common AI video generators.

Confirmed

• The company says JST-2 can generate complete 3D scenes within seconds and supports expanding scene boundaries in real time
• Characters come with motion capabilities out of the box, requiring no manual rigging, and animated characters can be created from a single image
• The model has a built-in physics engine, so environments carry physical properties rather than being merely pixel-level video
• Scene generation, characters, and motion are all handled by a single unified architecture

Unconfirmed

• @drcintas notes that real-world results such as 360° orbiting still need hands-on verification, as current information mainly comes from official promotional material

Why it matters

• With most world models trained in pixel/video space, JST-2's purely 3D-native approach, if it delivers in practice, could open a new technical direction for 3D content generation and interactive world modeling, while also eliminating manual rigging and other steps in traditional 3D production

2026-10-20 ~ 2026-10-20 · 6 related posts

• [source] Viggle releases JST-2, first full-stack physics-native 3D world foundation model — dr_cintas · 2026-10-20
• [source] Viggle unveils JST-2, an AI that builds physics-enabled 3D worlds in seconds — 141_1337 · 2026-10-20

4 near-duplicate retellings: Div_pradeep · Aiden_Tech_Ai · Aiden_Tech_Ai · anthara_ai