Cheapest image edit models on Sume: price per billed image

If your job is image-to-image at volume, price is a filter on top of the input_references descriptor. The catalog stores a static list price per image for each id, and Sume bills list times 1.25. Ranking the edit-capable ids on that number gives a short list.

These are catalog list prices, not quotes for your exact settings: some ids vary by tier or quality (Nano Banana 2 by resolution, Ideogram 4.5 by quality), and /v1/images/models/{id}/endpoints returns the current pricing array.

Edit-capable, cheapest first

Cheap is not the same as right

Pick on the job. For text-heavy layouts, Ideogram 4.5 costs more and prices by quality. For up to 16 references or a mask, ChatGPT Image 2.5 is the one with those fields. For a 10-reference product set, FLUX.2 Pro at $0.0375 billed is a middle option.

import os, requests
headers = {"Authorization": "Bearer " + os.environ["SUME_API_KEY"]}
url = "https://api.sume.com/v1/images/models/black-forest-labs/flux.2-pro/endpoints"
r = requests.get(url, headers=headers, timeout=30)
print(r.json()["endpoints"][0]["pricing"])

A cost check

A 1,000-image batch on Qwen Image is $25 billed at list; the same batch on ChatGPT Image 2.5 is about $65.90. Run a 20-image sample on each candidate and compare outputs next to the bill before you fix a model.

How this was checked

Vendor facts come from the pages listed in the sources, read on 2026-10-05. Sume facts come from the Image API docs and the catalog code on main on the same date. Catalogs and limits change, so read the descriptors from GET /v1/images/models before you pin a number in production code.