{
  "slug": "doctl-serverless-inference-endpoints-support-29ce85e42aa7",
  "title": "doctl serverless inference endpoints support",
  "tools": [
    "doctl"
  ],
  "tags": [
    "agent"
  ],
  "introduced_at": "2026-06-03T05:09:53.000Z",
  "announced_at": "2026-06-03T05:09:53.000Z",
  "effective_at": null,
  "date_basis": "source-publication",
  "status": "unknown",
  "how": {
    "incantation": "When developing or testing code that needs model inference, use doctl’s serverless inference commands to call the endpoint. For a chat prompt, run `doctl inference chat-completions create` with a model and message, or provide a full JSON request.",
    "examples": [
      "doctl inference chat-completions create --model <model-id> --message \"<message>\""
    ],
    "gotchas": [
      "Version: v1.160.1",
      "--model (required): Model ID; required unless `--request` is set. `-m` is an alias.",
      "--message (required): User message; required unless `--request` is set.",
      "--request: Path to a JSON request body; use `-` for stdin. When set, `--model` and `--message` are not required.",
      "--stream: Stream output using server-sent events.",
      "--access-token (required): Inference credential: a Model Access Key or a DigitalOcean PAT with full access scope.",
      "Prerequisite: Authenticate with `--access-token` using a Gradient Model Access Key or a DigitalOcean PAT with full access; limited scopes do not work for serverless inference.",
      "Inference APIs require a DigitalOcean PAT with full access scope or a Gradient Model Access Key; a limited-scope PAT results in 401 for inference calls.",
      "Docs conflict: The current DigitalOcean CLI reference documents `doctl serverless-inference`; the v1.160.1 source documents `doctl inference` and shows the chat-completions invocation. The current reference is generated from doctl v2.0.0, so its command naming is newer than the card version."
    ]
  },
  "why_it_matters": "Call a serverless inference endpoint while developing or testing code with doctl.",
  "sources": [
    "https://github.com/digitalocean/doctl/releases/tag/v1.160.1",
    "https://github.com/digitalocean/doctl/pull/1846",
    "https://github.com/digitalocean/doctl/blob/v1.160.1/commands/serverless_inference.go"
  ],
  "rank": 0.91022,
  "judge": {
    "publish": true,
    "judge_model": "codex-cli/gpt-6-luna+typesafe/jev-1.13.0",
    "prompt_version": "extract-v7+enrich-v2+judge-v2x3-rejudge+jev-publish-v1"
  },
  "review_state": "approved",
  "provenance": {
    "run_id": "wave2a-20261017",
    "verdict_id": "wave2a-20261017:wave2a-20261017-17c2f420bbac3b85-2:rescue",
    "evidence_hash": "bada8c4f164377d824857c50efa7d5e2c8891965432832049eb06ce08ea51a97"
  },
  "review": {
    "draft_hash": "af2b62574fdaec9dfcfb7c9702417aaebfd791e6beae63d8dcb749a588a10931",
    "reviewer": "publication-gate",
    "reviewed_at": "2026-10-07T00:00:00.000Z",
    "decision": "wave2a-20261017:rescue",
    "reason": "Rescue path: first rejected only because the extracted card lacked usage; judged again on the enriched card, 2 of 3 judge runs approved with p_yes >= 0.9 (mean 0.91 >= 0.9; 2 needed) and Jev gave 0.61 (>= 0.6 needed)."
  }
}