Deployments
A deployment publishes a trained adapter (or a base model) at a dedicated subdomain (https://<slug>.arkor.app/v1/chat/completions) that speaks the OpenAI Chat Completions wire format. Point any OpenAI-compatible client (the official openai SDK, LangChain, etc.) at it and it just works.
CloudApiClient is the typed entry point for managing these deployments programmatically (creating them, rotating API keys, deleting them) from a Node.js script or your own server. The Studio dashboard (Endpoints) covers the common create / toggle / key-management actions interactively; the SDK is the lower-level surface those tools sit on top of, and it’s what you reach for when:
- CI / IaC: provision deployments alongside your other infra. A trained job’s
finaladapter goes from “saved checkpoint” to “live URL” in one script step. - Custom dashboards: embed deployment management in your own admin UI.
- Bulk operations: script revoking + re-issuing keys across many deployments after an incident.
- SDK-only operations: re-targeting an existing deployment to a different job / checkpoint, configuring non-default run retention, etc. Studio doesn’t surface these today.
defaultArkorCloudApiUrl(credentials) keeps the client targeting the same staging / self-hosted control plane the user authenticated against: anonymous credentials carry that URL since signup, OAuth credentials since arkor login. The plaintext API key is also returned only on creation; subsequent listDeploymentKeys() calls return the label and a display prefix only.
Reference
For the full method list, request / response shapes, thetarget discriminated union, run-retention configuration, error codes, and the public-vs-internal export contract, see the CloudApiClient reference.