RenderDrama

What a second of AI video costs, channel by channel

What the step up in resolution actually costs

Every channel publishing more than one resolution publishes a ratio between them, and the ratios are wildly different: in the table below, one vendor charges the same for 720p and 1080p, another doubles. As of 2026-09-12.

Lower as a share of higher on the ledger, by model and tierEach bar is one row of the table below, drawn to its entry under “lower as a share of higher”. Nothing is plotted that the row itself does not state.Lower as a share of higher on the ledger, by model and tierVeo 3.1 Standard100%100%Veo 3.1 Fast83%83%Wan 3.050%50%
Fig. 1 Each bar is one row of the table below, drawn to its entry under “lower as a share of higher”. Nothing is plotted that the row itself does not state.
Resolution tiers, as published. Recorded 2026-09-12.
Model and tierLower tierHigher tierLower as a share of higher
Veo 3.1 Fast720p $0.10/s1080p $0.12/s83%
Veo 3.1 Standard720p $0.40/s1080p $0.40/s100%
Wan 3.0720P $0.10/s, 480P $0.051080P $0.20/s50%

Inclusion rule. Channels publishing a per-second rate at more than one resolution on the same page. A channel with a single rate has no ratio to record. Order. Alphabetical by model and tier.

1The ratio matters more than the rate

A production delivering at 1080p should compare 1080p rates, not headline figures. Channels advertise their lowest tier, and the step to a deliverable resolution costs nothing on one channel and doubles on another.

Comparing on the advertised figure therefore ranks channels in an order that does not survive contact with the delivery spec.

A close detail of hands repairing a wristwatch, carrying the pixel count that region would have in a 480-line frame, enlarged to the same size as its neighbours.The same detail carrying the pixel count it would have in a 720-line frame, enlarged to the same size.The same detail carrying the pixel count it would have in a 1080-line frame, enlarged to the same size.
Fig. 2 One image, resampled to the pixel count each rung would give this region and then enlarged to a common size. It shows what the rung buys in detail; it says nothing about how any particular model behaves at each rung.Source frame generated on 2026-09-16 and resampled here. Not output from any channel on the ledger, and not a comparison between channels.

2Where the step is free, take it

At least one tier here charges the same for both resolutions. There is no reason to generate the lower one except speed, and where that is true the lower tier serves mainly to make the headline rate look lower.

3Resolution is not all the tier changes

Tiers often differ in maximum clip length and in what references they accept as well as in pixels. A cheaper tier that also caps the clip shorter costs more per finished minute than the rate suggests, because it needs more calls and more joins.

4The ratio is a product decision, not a cost

Rendering at a higher resolution does not cost a vendor twice as much, so a doubled rate is pricing rather than physics. Reading it that way makes the comparison easier: the ratio says how a vendor wants the tiers used, not what they cost to serve.

Where the ratio is one, the vendor is effectively saying the lower tier exists for speed.

5Deliverable resolution should be chosen first

The comparison that matters is at the resolution the work ships at. Picking a channel on its cheapest tier and then discovering the delivery spec needs the tier above reverses the ranking often enough to be the standard mistake.

6Check what else moves with the tier

Clip length, accepted references and audio behaviour often change between tiers of the same model. A tier comparison limited to pixels per second misses the part that changes the shot list.

7Pages these figures were read from

A definition entry: what the term means, and the figures vendors put in public about it, each figure read on 2026-09-12. Related: Prompt extension, Per-video billing.

8Inside a generated second

Price aside, what goes into the seconds behind “Resolution tiers” is described in SceneMixer's guides:

The Seedance 2.0 tiers render segments up to 15 seconds; use a Seedance 2.5 or Wan tier for 30-second takes.

SceneMixer, Seedance 2.0 Fast — “Does Fast render 30-second takes?” (read 2026-10-13)

In the editor this tier is greyed out for such a segment; pick a Seedance 2.5 or Wan tier.

SceneMixer, Seedance 2.0 (standard tier) — “Can a segment longer than 15 seconds use this tier?” (read 2026-10-13)