sume/auto is missing from GET /v1/images/models: hardcode it

sume/auto is not listed by GET /v1/images/models, by design. The Sume docs call it a Sume-only model value: you send it in the model field of POST /v1/images and Sume selects the family for you. A model picker that is built only from the list will never show it, so add it yourself.

What the docs say

The Image API docs state three things about Auto. The catalog does not list it. The response keeps model: "sume/auto" and does not name the family that ran. Sume never discloses which family served the request. Image 1.0 is a compatibility alias for the same Auto pipe and is marked as retiring soon.

A picker that handles both

Fetch the list, then add the Auto entry by hand. This sketch uses the response shape from the docs, where data is an array of rows with an id.

import os
import requests

BASE = "https://api.sume.com"
HEAD = {"Authorization": "Bearer " + os.environ["SUME_API_KEY"]}


def image_model_ids():
    r = requests.get(BASE + "/v1/images/models", headers=HEAD, timeout=30)
    r.raise_for_status()
    ids = [row["id"] for row in r.json()["data"]]
    return ["sume/auto"] + ids


if __name__ == "__main__":
    for model_id in image_model_ids():
        print(model_id)

What you give up with Auto

Auto is convenient when you do not care which model draws. It is the wrong choice when you need repeatable looks, a known price or a feature that only some rows list.

• Price: you cannot read a per-image price for Auto from the list.
• Features: mask_url and background are listed on specific rows, so pin one for edits that need them.
• Audit: the job model stays sume/auto, so store your own metadata on each job.
• Drafts: a pinned model is the safer pick if a later final must match the draft.