1-minute vertical AI video ad: cost of four 15-second 9:16 clips

No Sume video row makes a 60-second clip in one request. The longest windows are 30 seconds, on Seedance 2.5 and Wan 3.0. A one-minute vertical ad is therefore two or more jobs joined afterwards, and the main cost question is which row, and how many clips, to use.

This post prices the two sensible layouts for a 9:16 ad: four 15-second clips, which fits every row except Gemini Omni Flash 1.1, or two 30-second clips, which only Seedance 2.5 and Wan 3.0 allow.

What do four 15-second clips cost?

Every row below lists 9:16. Each clip is its own job and is rounded up to the cent on its own, so the total is four times the per-clip figure.

What about two 30-second Wan clips?

Wan 3.0 accepts 30 seconds in one request and lists 9:16. At 480p one clip costs $1.88, at 720p $3.75 and at 1080p $7.50, so a full minute is twice that: $3.75, $7.50 or $15.00. Fewer joins means fewer visible cuts, which is the argument for the longer window.

What does the cut between clips need?

Continuity is the real cost. Generate each clip from the last frame of the one before: Sume's rows with last_frame and first-frame support accept frame images, and Kling, Wan and the MiniMax rows all take them. Join the clips on the Sume timeline or in your editor, then add the voice-over once across the whole minute instead of per clip.

For audio, Kling bills its two rates by the audio switch, while Wan and the MiniMax rows accept reference audio. Decide on audio before you generate; you cannot add it to a clip you already paid for at the audio-off rate without a rerun.