Zalando image size: ask for 2000x2880, an exact 1:1.44

Zalando asks for portrait images with an aspect ratio of 1:1.44, a best-practice size of 1801 x 2600 pixels, a minimum of 762 x 1100, JPG/JPEG, sRGB, and at most 20 MB. On Sume, image_size on POST /v1/images takes custom pixels only when both edges are multiples of 16, so 1801 x 2600 cannot be requested as written. 2000x2880 can: it is exactly 1:1.44, it clears Zalando's minimums, and it fits Sume's pixel box.

Zalando's side comes from its image guidelines page, which says it was updated on October 1, 2026 (read 2026-10-02). Sume's side comes from the Image API docs.

What exactly does Zalando require?

The page lists the baseline for every category. One detail matters for generated images: Zalando says its system rounds the ratio, and that a ratio of 1.445 or 1.446 becomes 1.45 and is marked as an incorrect aspect ratio. So a size that is only close can fail. Aim for exactly 1.44 and check the pixel count before you upload.

It also says that a correct file size with a blurry image leads to a rejection, so resolution has to be real, not an upscale of a thumbnail.

Which size should I send to Sume?

Sume's Image API docs say a custom image_size needs both edges to be multiples of 16, a maximum edge of 3840, an aspect ratio of at most 3:1, and 655,360 to 8,294,400 total pixels. 1:1.44 is 25:36, so exact sizes need a width that is a multiple of 400 once you also need multiples of 16. Two that work: 1600 x 2304 and 2000 x 2880. At 2000 x 2880 the image is 5,760,000 pixels, and both edges are multiples of 16 (125 x 16 and 180 x 16).

Both clear the 762 x 1100 minimum. Only 2000 x 2880 also clears the 1800 x 2600 minimum Zalando lists for designer brands, so use it if that applies to you. The size comes from the docs' rules; the model you pick still has to accept the field, so read its supported_parameters in GET /v1/images/models first.

PROMPT = "Keep the product exactly as in the reference. Plain white studio background, product centred, soft natural shadow, no text, no props."
import os, requests

r = requests.post(
    "https://api.sume.com/v1/images",
    headers={
        "Authorization": f"Bearer {os.environ['SUME_API_KEY']}",
        "Idempotency-Key": "zalando-sku-1042-primary",
    },
    json={
        "model": "openai/gpt-image-2.5",
        "prompt": PROMPT,
        "input_references": [
            {"type": "image_url", "image_url": {"url": "https://example.com/sku-1042.jpg"}}
        ],
        "image_size": "2000x2880",
        "output_format": "jpeg",
    },
    timeout=60,
)
if r.status_code == 200:
    print(r.json()["data"][0]["url"])
elif r.status_code == 202:
    print("poll", r.json()["data"]["status_url"])
else:
    r.raise_for_status()

What about the file type and the 20 MB cap?

Set output_format to jpeg; the docs list png, jpeg, webp and svg. A 2000 x 2880 JPEG is far below 20 MB in normal use, but check the file you actually get. The Sume docs I read do not say which colour profile the output carries, so confirm sRGB with your own tool before relying on it.

A blocking call waits up to 30 seconds. If the generation is still running, Sume returns 202 with a job envelope instead of the image, so check the status code, then poll the status URL as described in Jobs and results. Send an Idempotency-Key so a retry does not create a second paid job.

What should I do next?

Here is the short version, in the order to do it.

• Request 2000x2880 with output_format: "jpeg" for portrait Zalando slots.
• Check the real pixel size and ratio of every returned file before upload.
• Treat the ratio as a hard rule: close to 1.44 is not enough per Zalando's rounding note.
• Re-read Zalando's page before each season; it was updated on October 1, 2026.