Viggle AI is a Spatial GenAI platform with 45M+ users, backed by a16z. Our proprietary Joint-Space-Time (JST) model is physics-aware, built to create and control motion across characters, scenes and environments. It powers Viggle for controllable video generation, PINOC for 3D character animation, and Viggle Games, where those characters and motion run in real playable builds. Our mission is to make spatial content creation accessible and scalable.
This is the 3rd launch from Viggle AI. View more
Viggle JST-2
Launching today
JST-2 is Viggle’s physics-native world foundation model for generating 3D scenes, characters and full-body motion in one shared world. Scenes take about one second, characters come from one image or text with motion built in, and motion comes from text or video. Try it free in the PINOC browser app, or use the API or MCP with agents like Claude Code. Benchmarks are Viggle’s own blind human tests.
Free Options
Launch tags:Artificial Intelligence
Launch Team
Framer AI AgentsDesign and publish professional sites with AI
Promoted
JST-2 is Viggle’s physics-native world foundation model. It generates 3D scenes, characters and full-body motion in one shared 3D world.
Problem: Building 3D content is split into separate, slow jobs. Scenes, characters, rigging and animation each need their own tools and specialists, and rigging and skinning alone can take hours and decide how a character moves afterward.
Solution: JST-2 handles all three in one model. Viggle’s JST architecture represents the world as persistent 3D “atoms” with learned attributes and dynamics, encodes them as Physics-Native Tokens, and decodes them into images or 3D Gaussian splats.
What makes it different: Characters come with motion built in, so there’s no separate rigging or skinning step, and scenes can be explored without rerunning the model. Per Viggle, a 3D scene takes about one second to generate, against about 5 to 6 minutes for Marble, at an inference cost below $0.01 per scene. In Viggle’s own blind human tests, JST-2 was preferred over Meshy 7.1 and Tripo H3.1 for characters, and over HY-Motion-1.0 and DeepMotion Animate 3D for motion.
Key features:
Scenes: generate a 3D scene from prompts or images, expand a world one view at a time (up to 64 input views), or reconstruct real environments from sparse images for robot simulation
Characters: full-body 3D characters from one image or text, including cropped or partly hidden images, with detailed faces and individual finger movement
Motion: full-body 3D motion from a text sentence or from video, applied directly to JST-2 characters and compatible with standard 3D rigs
PINOC, a free browser app with Scene mode and Agent mode
API, plus MCP support for agents like Codex and Claude Code
Who it’s for: 3D artists, game and animation developers, and robotics and embodied-AI teams who want scenes, characters and motion without stitching separate tools together.
Try it free in the PINOC app, read the JST-2 launch page, and see the API docs.
Is PINOC free with no limit for now, or is there a cap once you start exporting characters for real projects?