YouTube channel profile picture: 98x98, 15 MB, formats, AI image API

YouTube's help page lists JPG, GIF, BMP or PNG under 15 MB, rendered at 98x98 px. Request a square PNG from Sume's image API and check it before you upload.

5 min readSume
All posts

A YouTube channel profile picture must be a JPG, GIF, BMP or PNG file (no animated GIFs), no larger than 15 MB, and an image that renders at 98 x 98 px, according to YouTube's Manage your channel branding help page, read on 2026-10-03. The picture also has to follow YouTube's Community Guidelines.

Sume's image API can produce that file: ask for a 1:1 image in PNG or JPEG and download it. Sume does not upload the picture to your channel; you add it yourself in YouTube Studio.

What does YouTube's help page require?

The page lists a short set of rules. Everything in the table below is from that page; the last row is a note about what the page does not say.

Channel profile picture rules from YouTube Help (read 2026-10-03)
ItemWhat the page says
File typesJPG, GIF, BMP or PNG; animated GIFs are not allowed
File sizeNot more than 15 MB
DimensionsAn image that renders at 98 x 98 px
ContentMust follow YouTube's Community Guidelines
Aspect ratioThe page I read gives no ratio, so a square source is a safe choice rather than a stated rule

Which file format should you ask the image API for?

Sume's Image API lists output_format values of png, jpeg, webp and svg, and a model only accepts the values its catalog entry lists. YouTube's profile picture list does not include WebP or SVG, so request png or jpeg and skip the other two for this use.

Ask for aspect_ratio: "1:1" so the picture is square. The docs show that Seedream 4.5 (bytedance-seed/seedream-4.5) accepts 1:1, and their endpoint example lists $0.033 per output image. Prices change, so read GET /v1/images/models/{model_id}/endpoints before you budget.

How do you make one with a single request?

Send POST /v1/images. The docs say the call blocks up to 30 seconds and returns 200 with data[].url; if the job runs longer you get 202 and a job envelope, so check the status code rather than the body shape. The sketch below handles only the quick 200 path.

Set n to get a few options in one call. The docs say you can request up to 10 per call but each model's ceiling is lower; the Seedream 4.5 example in the docs caps n at 4.

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": "Flat illustration of a friendly fox face, bold shapes, "
                  "centered, plain colored background",
        "aspect_ratio": "1:1",
        "output_format": "png",
        "n": 4,
    },
    timeout=60,
)
if resp.status_code != 200:
    raise SystemExit("Not finished yet: poll the job (status %s)" % resp.status_code)
for image in resp.json()["data"]:
    print(image["url"], image["media_type"])

What should you check before uploading?

YouTube states the picture renders at 98 x 98 px, which is small. These checks are my suggestions, not YouTube rules.

  • Shrink the result to a thumbnail-sized preview on your own screen: one simple subject survives that, fine detail does not.
  • Keep the file under 15 MB; a PNG of a flat illustration is normally far below that, but check the downloaded size.
  • Ask for no text or very little: lettering becomes unreadable at that size.
  • Generate more than one option and pick the one that stays clear at a glance.

How is a channel picture different from a thumbnail or a banner?

The three images have different jobs. The profile picture is the small mark that sits next to your name everywhere your channel appears, so it has to read at a glance and stay recognizable when it is tiny. A banner is wide art for the channel page, and a thumbnail sells one video. Because YouTube states the picture renders at 98 x 98 px, treat it like an app icon: one shape, strong contrast, a plain background.

When you write the prompt, describe the subject and the background separately, for example a centered fox face on a plain colored background, and name the style in a few words. Avoid asking for a scene. If a result has the subject off center, regenerate rather than cropping hard, since a tight crop of a small subject loses edge detail.

Where does Sume stop?

Sume does not change your channel. The upload and the crop happen in YouTube Studio. Sume's avatar product makes talking-presenter videos, not channel pictures, so for a still, use the image API.

If you start from a real photo, AI profile picture generator from a photo shows how to send it as a reference. For the wide image at the top of the channel page, see the channel banner sizes post.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume