How much does a 5-second AI video cost?

A 5-second AI video costs from cents to a dollar or two on Sume, depending on the model and the resolution. The table below lists the amount Sume reserves for a 5-second clip on each listed video model, before the 5.5% agent fee.

Sume bills routed video at the provider's list price × 1.25, as the Video generation docs say. Every amount below is computed from Sume's pricing code and rounded up to the cent, as read on 2026-09-29.

What does a 5-second clip cost on each model?

Most models price by the output second, so 5 seconds is five times the per-second rate. The Seedance models price by video tokens, which grow with width, height and length, so their amount depends on the frame too.

Can every video model make a 5-second clip?

Yes, on every model in the table. The docs list 4 to 30 seconds for seedance-2.5, 2 to 30 for wan-3.0, 5 to 15 for minimax-h3 and minimax-h3-max, and 3 to 10 for gemini-omni-flash-1.1; the seedance-2 sample lists 4 to 15. Current pricing code also accepts 3 to 15 seconds for kling-3 and 1 to 15 for grok-imagine-video-1.5, so five seconds fits every model in the table.

Check supported_durations on GET /v1/videos/models before you submit, because limits are not uniform across the catalog.

When is the money taken?

Sume reserves the estimate from the workspace balance when the request is accepted, adds a 5.5% agent fee by default, captures it when the job completes, and releases it if the job fails. The amounts in the table are the reserve, so they are what your balance must cover before you submit. If it cannot, the API answers 402 insufficient_credits.

Does sound change the price of a 5-second clip?

On kling-3 it does: the sound-on row costs more than the silent one, and sound is on unless you send generate_audio as false. On wan-3.0 the estimate is the same with sound on or off, and gemini-omni-flash-1.1 always makes sound, so its price already includes it.

How do I see the current rate before I submit?

Read pricing_skus on GET /v1/videos/models: it carries "pricing information per SKU" for each model (the docs sample shows a per-1,000-video-tokens rate for Seedance). See also the API pricing page. Provider list prices can change, so recheck before you budget a batch. For a longer piece, see How much does an AI video cost?.