Overview
The AIOZ AI API is the programmatic interface to the AIOZ AI Platform. It lets you run model inference, fetch model metadata and pricing, and check your account balance from your own applications, using the same compute network that powers the in-browser Playground.
What's available today
The API currently exposes the Inference surface across three resource groups:
- Tasks: Create an inference task, fetch its status and result, list past tasks, cancel a queued task.
- Models: Discover available models, fetch details, quote the cost of running a task, check whether a model is currently serving.
- Account: Check your API key's balance and earnings.
More resources (datasets, storage, organizations, and others) will be available in future releases.
Core concepts
- Task. An inference run on a specific model. Created with
tasks.create, polled withtasks.getuntil it reaches a terminal status. Tasks are asynchronous, see How Tasks Work for the full lifecycle. - Model. A runnable AI model on the platform. Each model has its own input/output schema, a price per inference, and a current serving status. Discover them with
models.listand fetch details withmodels.get. - Balance. Your account's available credits, free credits, debt, and accumulated earnings. Each successful inference draws from your balance, see How Billing Works.
API and Playground
The Playground is the no-code way to run a model. You fill in inputs in the browser and watch the result appear. The API does the same thing programmatically: every Playground run, under the hood, is an inference task on the same compute network. If you've tried a model in the Playground and want to ship it inside your own app, the API is the direct path.
Where to next
- Quickstart: get an API key, install an SDK, and run your first inference end-to-end.
- Authentication: how API keys work and how to send them.
- API Reference: the full list of endpoints.