YouTube community post images: 10 images, 16 MB, 1:1 and the API

A YouTube image post can hold up to 10 images, each a JPG, PNG, GIF or WEBP file up to 16 MB, and YouTube suggests a 1:1 ratio because that is how images appear in the feed, according to the Computer version of Create a post, read on 2026-10-03.

You can make a matching square set with Sume's Image API in one request. Posting still happens in YouTube; Sume only produces the files.

What are the image rules?

The help page I read gives these limits. Help pages can differ by device and app version, so check the composer on your own channel before you plan a large set.

How do you make a square set with the image API?

The Sume docs describe aspect_ratio values that include 1:1, and an output_format of png, jpeg or webp. Both png and webp are on YouTube's list, so either works.

The n field requests several images at once. The docs allow up to 10 per call but say per-model ceilings are lower, so read the n range in the catalog for the model you pick. If you need a set of 8 and your model stops at 4, send two requests with the same prompt and a small variation.

Keep a series coherent by writing one style sentence and repeating it in every prompt, then changing only the subject line. Sume has no seed control for images in v1: the docs say a seed parameter is rejected with 400 unsupported_parameter, so repeat the style sentence rather than expecting an identical look.

A request for a four-image set

This sketch prints the URLs for four options. It checks the status code first, because the docs say 200 is the image response and 202 is a job to poll.

import os
import requests

resp = requests.post(
    "https://api.sume.com/v1/images",
    headers={"Authorization": "Bearer " + os.environ["SUME_API_KEY"]},
    json={
        "model": "bytedance-seed/seedream-4.5",
        "prompt": "Poll announcement card, bold flat colors, big shapes, "
                  "no small text, subject: two coffee mugs",
        "aspect_ratio": "1:1",
        "output_format": "png",
        "n": 4,
    },
    timeout=90,
)
print(resp.status_code)
if resp.status_code == 200:
    for i, image in enumerate(resp.json()["data"], 1):
        print(i, image["url"])

What to watch for

• Text inside the image: a feed card is small, so keep any words short and large, and check them for spelling after generation.
• The expanded view: since viewers can click to see the whole image, a square with the key content in the middle is safer than one that relies on its edges.
• Disclosure: if an image shows realistic people or events, see the AI label rules for YouTube for the current checklist.
• Cost: each generated image is billed, so a 10-image set at the catalog price is ten image charges; check the price line first.

A simple workflow for a recurring series

If you post an image card every week, fix three things once: the aspect ratio (1:1), the output format (png) and the style sentence. Keep them in a small config in your own script and change only the subject line per post. Generate four options, pick one, and download it right away; keep the file that you uploaded so you can show what was posted later.

Treat each batch as a small spend. The docs' endpoint example shows a per-image price line, and the response carries a usage.cost value in USD, which you can log next to each post to see what a series costs over a month.

What Sume does not do here

Sume does not create the community post, add a poll to it, or schedule it. You paste or upload the files in YouTube. For wide art, see YouTube channel banner sizes, and for a small channel icon, see AI profile picture generator from a photo.