oiyai / docs

Welcome to Oiy AI

Compute for what comes next. Start with the tools and concepts you actually need.

Oiy AI brings GPU and CPU compute, persistent workspaces, and service lifecycle controls into one developer workflow. Bring a container, choose your resources, and build from the console or API.

DEVELOPER PREVIEWFeatures described here are implemented in the current product. Deployability depends on enabled regional capacity, pricing, and runtime readiness. The Python SDK, CLI, and MCP server currently use source distribution.

01 / Run your first service ↗

Move from a verified account to a configured workspace.

02 / Choose your compute ↗

Understand GPU profiles, CPU resources, and placement.

03 / Keep your work ↗

Learn what persists, what is billed, and what deletion removes.

04 / Build with your tools ↗

Use the HTTP API, Python SDK, CLI, or an MCP client.

How the pieces fit

A service is a container workload with a resource allocation, a persistent workspace, optional HTTP access, and a lifecycle you control. A volume stores the files you want to keep. A template is a reusable starting configuration.

You select a region tier (Standard or Low-price) separately from geography (North America, Oceania, or Automatic). The live catalog determines which combinations can actually be deployed.

A practical learning path

  1. Complete the quickstart.
  2. Read about service lifecycle, compute, and storage.
  3. Configure HTTP endpoints and idle sleep.
  4. Automate through the API, Python SDK, or MCP server.

Read the preview boundaries before planning a production dependency.

Your first workspace

Create a service, check its state, and keep the work you want to retain.

On this page

How the pieces fitA practical learning path