One finished minute of AI video with voice, music and captions: cost

A finished 60-second video built from six 10-second Wan 3.0 clips at 720p costs $7.98 on Sume once you add narration, a music track, a Timeline render and captions. The add-ons come to $0.48, so footage is more than 90 percent of the bill. Swap in Seedance 2.5 and the same minute costs $35.16.

The add-ons, job by job

Each job is rounded up to the next cent, which is why the totals are not the exact products of the rates.

Footage by model

Six 10-second clips at 720p, rounded per clip. Seedance is token-priced, so its per-second figure at 16:9 720p is $0.378 for Seedance 2 and $0.5778 for Seedance 2.5.

What this does not include

It assumes every clip is kept. Retakes add whole clips, and a retake rate of one in four adds a quarter to the footage line. It also assumes a minute of narration is 900 characters; your script may be longer or shorter, but narration is cheap either way.

The choice that moves the bill is the model. For a related run at 1080p with regeneration overhead, see cost per finished minute at 1080p, and for a comparison with Google's rates see cost per minute against Veo 3.1.

Two cost choices matter more than any add-on. First, clip length: six 10-second clips and three 20-second clips of the same model cost nearly the same, but fewer, longer jobs are easier to manage and each job rounds up only once. Second, the length of narration: TTS at $0.0475 per 1,000 characters is so low that a script twice as long adds only a few cents. The Timeline render is also priced per output minute, so a 60-second finish is $0.10, and the captions job is flat for videos up to 60 seconds, which is exactly the length this example uses.