Sume image edit returns 400: input_references on a text-only model

An edit that sends input_references to a model whose catalog descriptor is {"min": 0, "max": 0} is rejected: the Image API docs say such models are text-to-image only and reject references. The fix is to read the descriptor and choose a model whose input_references max is above zero, or to drop the references and generate from text.

What the docs say

From the Image API docs.

Find models that take references

List the catalog and keep those with a reference max above zero.

import os, requests

r = requests.get(
    "https://api.sume.com/v1/images/models",
    headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
    timeout=30,
)
for m in r.json()["data"]:
    ref = m.get("supported_parameters", {}).get("input_references")
    if ref and ref.get("max", 0) > 0:
        print(m["id"], ref["max"])

Other causes of the same error

unsupported_parameter is a general result for anything the model does not list, so also check mask_url (documented for ChatGPT Image 2.5), background, quality values and seed, which no model serves in v1.

Prevent it

Run the descriptor check in CI against your saved request templates, so a model swap fails a test instead of a customer request.