Make images, video and sound where you design
Open Generate in the canvas toolbar, pick a model and describe what you want. The results land on the canvas as layers.
Models in the Generate panel
Nano Banana Pro
GPT Image 2.5
Seedream 5.0 Pro
FLUX 3
Recraft V4.1
Kling 3.0 Pro
Wan 2.7
Eleven V4
From prompt to layer
Reference images
Click images on the canvas to use as references, up to the number each model accepts.
Variations
Make one to four images from a single prompt. They land together on the canvas, ready to compare.
Video and audio
Turn text, an image or references into video. Generate speech, change a voice or make sound effects.
Image tools
Remove a background, cut objects out into their own layers, vectorize, crop and annotate.
Vector graphics
Switch to Vector to generate icons and illustrations as shapes you can edit.
Provider cost
Runs draw from your workspace balance at provider cost, rounded up to the cent. Runs on your own API keys are never billed here.
Related
- ProductsBuilder Design AgentBuilder, the agent that designs, checks and publishes pages on your canvas.
- GlossaryWorkflowThe part of a canvas that makes material for your pages without publishing itself: prompts, documents, and the images, video and audio generated from them.
- ProductsBuilder DesignThe canvas: auto layout, components, variables and live multiplayer.