API predictions are currently free and use no credits, but creating them needs an active Infinite plan. Without it, the API answers 403 plan_required, while reading, listing and canceling your predictions keeps working.

Four: Seedance 2.5 Lite and PicassoIA Video for text and image to video, PicassoIA Image for text to image, and PicassoIA Image Editor Pro to edit or combine 1 to 4 images.

Up to 5 per account, queued or running, across all its API keys and shared with the generations it runs through MCP (ChatGPT, Claude, Supercomputer). The 6th answers 429 concurrency_limit.

Create it in the API keys section of your account. The full key is shown only once, an account can hold 2 keys at a time, and keys belong on your server: the API sends no CORS headers, so browsers cannot call it.

Create a prediction with a POST, poll it until its status is succeeded, failed or canceled, then read output: a list of image URLs for the image models or one MP4 URL for the video models.

cURL, Node.js and Python. Each model page also has a prompt for your coding agent and a button to try the model on Picasso IA.