Image-to-video shot: a 2-cent still plus a 5 s clip is 65 cents

The short answer

A still made with GPT Image 2.5 at medium quality costs 2 cents on Sume. Animate it as a 5-second 720p clip and the shot costs 65 cents on Wan 3.0 (2 + 63) or 291 cents on Seedance 2.5 (2 + 289). A hundred shots cost $65.00 or $291.00. The still is about 3 percent of the Wan shot.

Shot cost by clip model

Sume bills the provider list price times 1.25, rounded up to whole cents for each job (the catalog formula reads "list x 1.25, ceil usd cents"). The amounts below come from that formula, applied to the catalog as of 2026-10-08.

Why start from a still

An image-to-video request takes an image_url (and optionally an end frame) and animates it, so you can judge composition for 1 to 2 cents before paying for motion. Ten stills at medium quality cost 20 cents. If you generate ten and animate the best two on Seedance 2.5, the total is 10 x 2 + 2 x 289 = 598 cents, against 2890 cents for ten Seedance 2.5 text-to-video clips.

The image models work in 1024-class sizes at the quoted rates (low 1 cent, medium 2, high 7). Match the still to the clip you plan to make, because the video request carries its own aspect_ratio field.

Choosing the still quality

Medium (2 cents) is a middle choice. A low still is 1 cent, a high still is 7. On Wan 3.0 the shot would be 64 cents with a low still, 65 with medium and 70 with high, so the still tier moves the total by under 10 percent. Because the still sets the look that the clip then moves, many teams spend more on it than on the first test of the motion.

For a 20-shot storyboard on Wan 3.0 at medium, the stills cost 20 x $0.02 = $0.40 and the clips 20 x $0.63 = $12.60, a total of $13.00.

Caveats

This is the generation cost only. A still you reject saves the clip you would have run, but a still you accept and then animate badly still costs its clip. The 5-second length sits within every model listed: Wan 3.0 accepts 2 to 30 seconds, Seedance 2.5 4 to 30, and Kling 3 4 to 15. See Images for image quality tiers and Video Router for the image_url field.