Supported
Generation
Start generations, estimate their credit cost before running, and check status and history.
Assets
List, upload, import, fetch, download, and delete generated and uploaded media.
Model discovery
List models, inspect their parameters, and optimize prompts.
Account
Read credit balance, ledger, plan, and API limits.
Workflows
List workflows already saved in Pixio and run them.
Boards & Canvas
Full project CRUD and operations for Boards, spatial canvas, and Canvas designs.
Cinema & Cam View
Storyboard CRUD with prompt-to-storyboard direction; Cam View scene CRUD with prompt-to-keyed-scene direction.
Agent & timeline projects
Persist chat and Video Agent projects; timeline project CRUD with validated editor operations.
Characters, prompts & training
Locked character CRUD, prompt history, and model-training status.
Not supported
These are deliberately app-only. Don’t design around them.The pattern: authoring and reading are open, long-running orchestration and live device links are not. You can construct the work, then finish it in the app.
Planning around the gaps
I want a fully automated video pipeline
I want a fully automated video pipeline
Generate clips through the API and assemble them yourself, or build the timeline via the API and do the final render in the app. Headless render isn’t exposed.
I want to train a LoRA on a schedule
I want to train a LoRA on a schedule
Not possible today — training starts in the app. You can poll training status and react when one finishes.
I want my own chat UI over Pixio
I want my own chat UI over Pixio
Build it on the generation endpoints directly rather than the internal chat route. You get the same models without depending on an unsupported surface.
I want the same chain every time
I want the same chain every time
Build it once as a workflow in the app, then call it by ID. Cheaper than reimplementing the chain in your own code, and it stays in sync when you edit it.
Reference
Every supported operation has its own page with parameters and a live playground, generated from the API’s own specifications:GET /capabilities returns the supported surfaces programmatically, so an integration can check rather than assume.