AliExpress product image size: 1000px square, then what to ask Sume

What size should an AliExpress product image be?

AliExpress's Image Size Guide says to use a minimum resolution of 1000x1000 pixels for product listings, keep the image under about 2 MB, and prefer a white background. It also says to set your editor to 1080x1080 px at a 1:1 ratio for all products, and it shows 4:5 as the ratio for apparel such as dresses in its text.

One caveat before you build a pipeline on these numbers: this is a general AliExpress guide page, not the seller-center upload form, and AliExpress's own pages are not identical. Another AliExpress article I read, How to Add Pictures to Your AliExpress Product Listings, asks for at least 1200x1200 px and 1500x1500 px for zoom detail shots. Aim for the larger figure and confirm in your own seller center, which I could not open.

What the AliExpress guide actually lists

The guide's comparison table lists dresses, key rings and roof racks. Each row gives 1080x1080 px, 1:1 and 150 PPI. The background column reads white or light gray for dresses and roof racks, and a neutral tone for key rings. The text separately says images smaller than 800x800 px may look blurry on high-definition screens, and that 150 PPI is the minimum resolution to export at.

Format is only mentioned in passing: the guide suggests JPEG or WebP for fast loading. It does not give a limit on the number of images in the pages I read, so this post does not state one.

What to ask Sume for

Use the Image API with openai/gpt-image-2.5. Per the Image API docs, it accepts background: auto|transparent|opaque and quality: auto|low|medium|high|xhigh|max, with high as the default when you omit it. It takes up to 16 input references, so you can pass the supplier photo and ask for a clean white-background rework.

The exact 1080x1080 canvas is the awkward part. For custom pixels on GPT models, both edges must be multiples of 16, and 1080 is not. The nearest legal square at or above 1080 is 1088x1088, which sits inside the documented limits (maximum edge 3840, aspect at most 3:1, 655,360 to 8,294,400 pixels in total). Ask for 1088x1088 and downscale to 1080 in your own export step, or leave it at 1088 if the upload form accepts it. For apparel at 4:5, Sume's aspect catalog notes 1024x1280 as the legal way to get a 4:5 canvas on GPT models. That is 1024 by 1280, so it clears the 1000 px minimum on the short edge.

Compute your own output size before you upload: the 2 MB guidance is a file-size figure, and Sume does not promise any file size. The docs list output_format as png, jpeg, webp or svg, so request jpeg or webp and check the size afterward.

A request you can run

This sends a supplier photo through an edit with an opaque background and four candidates. The docs say n goes up to 10 per call with per-model ceilings below that, so read the n range in the catalog before raising it. The response is a 200 with data[].url, or a 202 job envelope for slow configurations, so check the status code.

import os
import requests

resp = requests.post(
    "https://api.sume.com/v1/images",
    headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
    json={
        "model": "openai/gpt-image-2.5",
        "prompt": "Keep the product identical; place it centered on a clean white background, soft even light, no text or watermark",
        "input_references": [
            {"type": "image_url", "image_url": {"url": "https://example.com/supplier-photo.jpg"}}
        ],
        "image_size": "1088x1088",
        "quality": "high",
        "background": "opaque",
        "output_format": "jpeg",
        "n": 4,
    },
    timeout=60,
)
print(resp.status_code)
if resp.status_code == 200:
    for image in resp.json()["data"]:
        print(image["url"])
else:
    print(resp.json())

Check before you publish

Open the result at full size and confirm the product edges, the label text and the colors match the real item. The AliExpress guide asks for a clean background and no distortion, and an edit model can alter small print on packaging. If a model rejects a field, Sume returns 400 unsupported_parameter rather than ignoring it, so a rejected request tells you which setting that model does not list. Then confirm the final dimensions, PPI and file size against your seller center before uploading.