YouTube first-frame views: check your opening frame at t=0

YouTube says views are now counted from the first frame, so the opening frame is what a feed or hover viewer sees the moment a view is recorded. To inspect it, submit your clip to Sume POST /v1/video-frames with at set to [0, 0.5, 1] and read the returned stills.

YouTube facts are from its blog post, read 2026-10-01; Sume facts are from the Video frames docs. This is a way to look at your video, not a way to change how YouTube counts.

What changed in how YouTube counts a view?

The post says that starting August 24, 2026, views across Shorts, long form, podcasts and live are counted from the first frame. The previous definition is now called an engaged view: clicking to watch, or watching past the initial seconds. A thumbnail you hover over so it autoplays counts as a view from the first frame, while an untouched thumbnail stays a thumbnail impression. The post also says analytics such as CTR, AVD and retention, recommendations and the Partner Program are unchanged. Compare with how YouTube counts Shorts views.

How do I pull the opening frames?

Import the clip to media.sume.com first, since video frames takes one workspace-hosted clip and no open-internet URL. Send exactly one of at[] or fps. Submit always returns 202, so poll GET /v1/video-frames/{id} until resource_status is ready.

const headers = {
  Authorization: "Bearer " + process.env.SUME_API_KEY,
  "Content-Type": "application/json",
  "Idempotency-Key": "first-frames-001",
};
const sub = await fetch("https://api.sume.com/v1/video-frames", {
  method: "POST",
  headers,
  body: JSON.stringify({
    video_url: "https://media.sume.com/artifacts/artf_demo/talk.mp4",
    at: [0, 0.5, 1],
  }),
});
const { request_id } = await sub.json();
console.log("poll /v1/video-frames/" + request_id);

What should I look for in those stills?

A silent autoplay viewer or a hover preview sees the first second without a deliberate click, so ask whether frame 0 already shows the subject, readable text and the framing you want. A black or logo-only frame at 0 is what that viewer gets.

What goes wrong?

An at value outside the clip fails with frame_time_out_of_range; see frame_time_out_of_range. Frames are extracted at source size unless you set max_edge (16 to 2160).