doctl dedicated-inference endpoints and tokens commands

Introduced 2026-03-31 · score 0.90

HOW TO USE IT

When managing a dedicated inference endpoint from a repository workflow, use the `doctl dedicated-inference` command group. Update it with a JSON or YAML spec using `--spec` (use `-` for stdin), and optionally pass a Hugging Face token for gated models.

WHY THIS HELPS

Create, inspect, list, update, or manage a dedicated inference endpoint from a repository workflow.

REQUIREMENTS AND LIMITATIONS

• Version: v1.154.0
• <dedicated-inference-id> (required): ID of the endpoint to update.
• --spec (required): Path to a JSON or YAML spec; `-` reads it from stdin.
• --hugging-face-token: Hugging Face token for accessing gated models.
• --format: Comma-separated output columns. Allowed: `ID`, `Name`, `Region`, `Status`, `VPCUUID`, `PublicEndpoint`, `PrivateEndpoint`, `CreatedAt`, `UpdatedAt`.
• Prerequisite: Authenticate `doctl` with a DigitalOcean API V2 access token.
• Update requires a dedicated inference ID argument; the command returns an error when it is missing.
• Docs conflict: Current CLI reference is generated from doctl v1.169.0, later than the card's v1.154.0. It documents the same update interface, but does not establish the v1.154.0 behavior for every operation in the card.

FIRST-PARTY EVIDENCE

• https://github.com/digitalocean/doctl/releases/tag/v1.154.0
• https://github.com/digitalocean/doctl/pull/1825
• https://docs.digitalocean.com/reference/doctl/reference/dedicated-inference/update/

Check the installed version and the source before using it. Entry JSON

Report a problem with this card