Qwen Image vs Qwen Image Max on the Sume image API

Sume lists two Qwen image models. qwen/qwen-image is $0.025 per image and does text-to-image plus reference-guided editing with up to 10 references. qwen/qwen-image-max is $0.09375 per image, 3.75 times as much, and is text-to-image only: its input_references range is 0 to 0. If a workflow edits a photo, only the cheaper row can do it on Sume.

Data below is from the Sume catalog (GET /v1/images/models), read 2026-10-03, and the Image API page.

The two rows

Both rows share the same 13 aspect ratios, which include 21:9, 9:21, 1:2 and 2:1, and the same n range of 1 to 4.

What happens if you send a reference to Max

The catalog describes Max as text-to-image only, and a request that sets a parameter the model does not list is rejected with 400 unsupported_parameter. The model's input_references descriptor is {"min": 0, "max": 0}, which the docs define as text-only: references are rejected. Check the descriptor in code before you route:

import os
import requests

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

Cost of picking Max

A hundred images cost $2.50 on Qwen Image and $9.375 on Max. The listed price is a flat per-image number for both, with no tiers in the catalog. The catalog does not state why Max costs more, and Sume publishes no quality ranking between them, so decide with a side-by-side on your own prompts: run each id with n: 4 for about $0.10 and $0.375. A call that is still running after the 30-second wait returns 202 with a job envelope instead of 200, so check the status code before reading data; see Jobs and results.