OpenAI

Who is it for?

OpenAI is a cloud-provider option for a product release that explicitly enables it.

Prerequisites

Use an approved agent or product settings flow and a credential you are authorized to use. Never paste it into a URL, chat, issue, email, or source repository.

Check readiness

Use the approved status view. A prepared contract or adapter does not make a cloud provider available in every product.

Choose a model

Choose only from models exposed by the released product flow. Model availability and cost are not promised by this guide.

Limits

Cloud access requires a released product capability and authorized credentials. Flow workflow inference is not available. After the matching release, the separate Provider Preview may issue one explicit neutral request capped at 32 output tokens; it remains invite-only, is not a workflow run, and requires entitlement, product-specific pairing, and a local OpenAI grant.

Cancellation and errors

Use the product's cancel action and its non-sensitive status. Do not create custom endpoints or credential forwarding paths.

Privacy and credentials

Store credentials only in approved settings. Review the applicable product privacy information before use.

Compatibility

See agent compatibility and provider overview.

Troubleshooting

Verify readiness and authorization through the approved UI. Report error context without secrets or customer data.