Cheapest 10-second 1080p AI video on Sume: five models ranked
A 10-second 1080p clip costs $1.40 on Kling 3 with audio off and $2.50 on Wan 3.0. The ranking by Sume price, with what each model does at that size.

The cheapest way to get a 10-second 1080p clip from Sume's Video Router is kling-3 with audio off at $1.40, then gemini-omni-flash-1.1 at $1.88, minimax-h3-max at $2.00, kling-3 with audio on at $2.10, and wan-3.0 at $2.50. Every price is the catalog's provider list rate times 1.25, rounded up to the cent for the whole clip, as the Video Router docs describe.
Seedance models also reach 1080p, but they are priced per thousand video tokens rather than per second, so they cannot be ranked from a per-second figure and are left out of this table.
What is the ranking?
Models without a 1080p tier are not in the list: minimax-h3 stops at 768p, and grok-imagine-video-1.5 stops at 720p.
| Model id | Resolution | Sume price per second | 10-second clip | Length note |
|---|---|---|---|---|
| kling-3 (audio off) | 720p or 1080p | $0.14 | $1.40 | 10 s |
| gemini-omni-flash-1.1 | 1080p | $0.1875 | $1.88 | 10 s (its maximum) |
| minimax-h3-max | 1080p | $0.20 | $2.00 | 10 s |
| kling-3 (audio on) | 720p or 1080p | $0.21 | $2.10 | 10 s |
| wan-3.0 | 1080p | $0.25 | $2.50 | 10 s |
Is the cheapest one the right one?
Not automatically. Kling 3 lists 16:9, 9:16 and 1:1 only, takes no reference images, videos or audio, and its rate is the same at 720p and 1080p, so on Kling the 1080p setting costs you nothing extra. That makes Kling the natural choice when you want the largest picture for the least money and your brief is a plain text or first-frame prompt.
Gemini Omni Flash 1.1 is the second cheapest, but 10 seconds is its ceiling, it lists only 16:9 and 9:16, and its sound is always on. Its 1080p rate is $0.1875 per second, which is why the 10-second clip rounds up to $1.88.
What do the dearer models add?
H3 Max is $0.20 per second at 1080p and accepts reference images, videos and audio, with limits of 9, 3 and 3. Wan 3.0 is the most expensive here at $0.25 per second, but it also accepts references, goes up to 30 seconds, and starts at 2 seconds. Of these five, only Wan reaches 2 seconds, and only Wan goes past 15, so the ranking above is not the whole decision.
For a draft-then-final workflow the gap matters more than the single price. A 10-second Wan clip at 480p is $0.63, while the 1080p final is $2.50; the tier is where the money is, not the model.
What if I only need 5 seconds?
The order does not change, because every row is linear in seconds, but the sums halve: $0.70 for Kling with audio off, $0.94 for Omni, $1.00 for H3 Max, $1.05 for Kling with audio on and $1.25 for Wan 3.0. Rounding up to the cent matters more on short clips: Omni's 1080p rate of $0.1875 per second gives $0.94 for 5 seconds, not the $0.94 a straight halving of the 10-second price would suggest.
If you will pick one clip out of several takes, the arithmetic changes again. Four takes on Kling 3 with audio off cost $5.60, which is still less than one 10-second Wan clip at $2.50. That is the case for starting with the cheap row and switching only when the brief demands references or length.
What about Seedance at 1080p?
Seedance 2.0, 2.0 Fast, 2.0 Mini and 2.5 all list 1080p in the Video Router catalog, with 4 to 15 seconds for the 2.0 family and 4 to 30 seconds for 2.5. They are billed by video tokens: the catalog's pricing basis for those rows is per 1,000 video tokens, and the per-clip price depends on the token count the provider reports, not on a flat rate per second. That is why they are not in the table, not because they are out of the running.
To price a Seedance clip, read the model's pricing block from the catalog and the usage the job reports. Do not extrapolate from the per-second rows above.
How current is this?
Rates are copied from the catalog. The Wan line in the catalog is dated 2026-08-24 and the H3 Max line 2026-09-08, and both are marked subject to change, so read GET /v1/video-router/models for the live pricing block before you commit a batch. The reserve placed on submit uses the same rule, so a clip never costs more than its hold.
Sources
Related posts
More in Pricing
- Cheapest Sume image models that take a reference photo
Reference edits start at $0.025 per image on Sume. Ranked list of every catalog model that accepts input_references, with reference limits and prices.
- Opus 5.5 costs 20% less than Opus 5: video agent turn math
Opus 5.5 lists $4 input and $20 output per MTok against Opus 5 at $5 and $25. Priced on a 200K-token video-agent turn, and what Sume did with Opus 5.
- Colossyan Professional $59: cost per minute of NEO and NEO2 video
Colossyan Professional lists $59 a month for 30 NEO minutes and 10 NEO2 minutes: $1.97 per NEO minute, or $1.48 per minute if you use every minute of both.
- How much do 1,000 AI images cost on the Sume image API?
Per-image list prices from the Sume image catalog for 18 model rows, multiplied out to 1,000 images, with the rows where the listed price is only a default.
Written by Sume