Log these six fields for every Sume image call

For every image call, log six things: the job id, Sume's request id, your idempotency key, the model id, the quality, and a metadata tag. With those you can recover a result after a crash, retry safely, and give support what it needs.

Models change often, as Higgsfield's changelog shows with FLUX 3 Image on 2026-10-01 (Higgsfield changelog, read 2026-10-02). Logging the model id and quality per call lets you compare outputs when you switch.

Why these six?

Each answers a question you will ask later. Sume puts the request id in the error body and in response headers, and its docs say to include it when reporting an issue, never an API key or signed URL.

What does the helper look like?

One function wraps the call and prints a JSON line. The metadata object is stored on the job, so use it for your campaign or SKU tag.

import json
import os
import requests

URL = "https://api.sume.com/v1/images"
HEADERS = {"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"}


def generate(prompt: str, key: str, tag: str) -> dict:
    body = {"model": "sume/auto", "prompt": prompt, "mode": "async",
            "quality": "medium", "metadata": {"campaign": tag}}
    r = requests.post(URL, json=body, timeout=60,
                      headers={**HEADERS, "Idempotency-Key": key})
    payload = r.json()
    print(json.dumps({
        "status": r.status_code,
        "request_id": r.headers.get("x-sume-request-id"),
        "idempotency_key": key,
        "model": body["model"],
        "quality": body["quality"],
        "campaign": tag,
        "job_id": (payload.get("data") or {}).get("job", {}).get("id"),
    }))
    return payload


generate("Studio shot of a ceramic mug", "mug-001", "autumn-launch")

What should I never log?

Never log API keys, signed URLs, or raw media URLs where others can read them. Store the Sume-hosted URL from the result, not provider URLs. The Errors docs describe the request id, and the Image API docs describe metadata.