YouTube thumbnail sizes fhd, qhd and uhd: frames from a clip

YouTube's Data API revision history says that on September 11, 2026 thumbnail definitions gained "fhd" (1080p), "qhd" (1440p) and "uhd" (4K) sizes. That is the API returning larger thumbnails; it does not say what you must upload, and Sume's video-frames route can produce stills up to a 2160 long edge or at the source's own size.

What does the revision note say?

The note names three new keys and the resources whose thumbnail definitions were updated: activities, liveBroadcasts, playlistItems, playlists, search and videos. It quotes resolutions in words (1080p, 1440p, 4K) and not in pixel pairs, so no pixel dimensions are asserted here.

Can Sume make a frame at each size?

max_edge is a long-edge clamp from 16 to 2160. Omit it and the frame keeps the source size, which the docs call the restage path. So a still cannot be enlarged past its source: if a clip is smaller than a key's resolution, the still is smaller too.

Ask for the same instant in several sizes by submitting one request per max_edge value; each request takes 1 to 24 at[] times.

curl -X POST https://api.sume.com/v1/video-frames \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: thumb-1080-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/clip.mp4",
    "at": [3.5],
    "format": "jpeg",
    "max_edge": 1920
  }'

Should I generate a thumbnail from a frame at all?

A frame is exact footage, which suits a thumbnail that must match the video. A designed cover is a different job; see YouTube vertical thumbnail size for that route.

What are the limits of this answer?

The revision note is about response sizes, so the choice of upload size stays with you and YouTube's current help page. Sume's route needs the clip to be a Sume-hosted media.sume.com URL (import first) and no longer than 300 seconds. A frame whose extract fails returns url null without failing the job.

How do I pick a size in practice?

Start from the clip you actually have. Extract one still first and read its width and height from frames[], then decide whether to keep the source size or clamp it. A clamp is the only size control on the frames route, so the practical choices are two: the source size, or a long edge you name between 16 and 2160.

If you plan to call the YouTube API afterward, read the thumbnail keys it returns for your own video and compare them with the still you made, instead of assuming a fixed pair of numbers.

• Keep the source size when the clip is already at the resolution you want to show.
• Use max_edge when a smaller file is easier to review or store.
• Use png when you want to inspect the frame without lossy artifacts.
• Repeat the request per size; there is no multi-size option in one call.