Generate a scene
curl --request POST \
--url https://api.gizmo.antimlabs.com/v1/scenes \
--header 'Content-Type: application/json' \
--data '
{
"prompt": "<string>",
"model": "<string>",
"asset_pipeline": "auto",
"persist": true
}
'{
"ok": true,
"scene_id": "jh72k3m4n5p6q7r8",
"job_id": "job_a1b2c3d4",
"status": "queued"
}Scenes
Generate a scene
Start async scene generation from a text prompt. Returns immediately with a job_id.
Poll progress via GET /v1/jobs/{job_id} or stream real-time events via GET /v1/jobs/{job_id}/events.
The AI agent will:
- Plan reusable assets from your prompt
- Generate each asset (geometry, joints, materials, physics)
- Create a floorplan and place assets with constraint solving
- Run physics validation
Typical generation time: 2-5 minutes depending on scene complexity.
POST
/
v1
/
scenes
Generate a scene
curl --request POST \
--url https://api.gizmo.antimlabs.com/v1/scenes \
--header 'Content-Type: application/json' \
--data '
{
"prompt": "<string>",
"model": "<string>",
"asset_pipeline": "auto",
"persist": true
}
'{
"ok": true,
"scene_id": "jh72k3m4n5p6q7r8",
"job_id": "job_a1b2c3d4",
"status": "queued"
}Headers
Body
application/json
Request body for scene generation.
Natural language description of the scene to generate
Minimum string length:
1Example:
"A modern robotics lab with two workbenches and a mobile manipulator"
Override the default LLM model
Asset geometry pipeline: 'auto' (recommended), 'gizmo' (parametric), or 'cad' (B-Rep)
Available options:
auto, gizmo, cad Whether to save the scene to your library
Response
Job created
⌘I