In Nutanix Enterprise AI, an API key is used to authenticate and authorize access to a specific inference endpoint. The entire lifecycle of an API key is tied to an endpoint that serves a Large Language Model (LLM). Therefore, an endpoint must be created and running before the system allows for the generation of an API key to access it. The user interface workflow requires selecting an existing endpoint as the target for the new API key.

A. Download LLM and create endpoint: While creating an endpoint is necessary, this option is less precise. Downloading an LLM is a prerequisite for creating an endpoint, not directly for creating the API key itself.

B. Provision an API key in NKP: API keys for model inference are managed within the Nutanix Enterprise AI application layer, not in the underlying Nutanix Kubernetes Platform (NKP).

C. Provision a GPU worker node: This is a cluster infrastructure requirement for running models, not a direct software prerequisite for the administrative task of creating an API key within the application.

1. Nutanix Enterprise AI User Guide v1.0: In the "Managing API Keys" section, the prerequisite is explicitly stated: "Before you begin: Create an endpoint." The procedure then details selecting an endpoint from the Endpoints page before the "Create API Key" option is available. (Reference: Nutanix Enterprise AI User Guide, Chapter: "Managing Endpoints and API Keys", Section: "Managing API Keys").