720p, 10 s, 9:16 on Sume: seven rows qualify, from $1.25 to $5.78
Which Sume video rows accept a 720p, 10-second vertical clip, which launch-week rows do not, and what each row bills.

Seven Sume rows can make a 720p, 10-second, 9:16 clip: the four Seedance rows, Wan 3.0, Gemini Omni Flash 1.1 and Kling Video v3 Pro, which renders 1080p instead. The billed price ranges from $1.25 (Wan 3.0 or Omni Flash) to $5.78 (Seedance 2.5) for the 10 seconds.
This is a useful filter in a launch week, because new rows from other vendors often add a resolution or ratio that Sume's rows do not share.
The rows that qualify
A 720p request must be a listed resolution (or Kling's alias), 10 seconds must be in range, and 9:16 must be allowed. Seedance prices come from the token estimator in the repo for a 10-second 9:16 clip, with no reference video, times 1.25.
| Row | Billed for 10 s | Notes |
|---|---|---|
| Seedance 2.5 | $5.78 | 4 to 30 s |
| Seedance 2.0 | $3.78 | 4 to 15 s |
| Seedance 2.0 Fast | $3.02 | 720p is its top |
| Seedance 2.0 Mini | $1.89 | 720p is its top |
| Wan 3.0 | $1.25 | 2 to 30 s |
| Gemini Omni Flash 1.1 | $1.25 | 3 to 10 s, audio always on |
| Kling Video v3 Pro | $1.40 (audio off), $2.10 (on) | Renders 1080p |
The rows that do not
These rows fail at least one of the three tests:
- Veo 3.1 (three rows): 8 seconds is the longest.
- MiniMax H3 and H3 Max: no 720p; they list 480p and 768p (H3 Max also 1080p).
- Grok Imagine Video 1.5: no
aspect_ratiofield; the clip follows the still. - H3 Max Recast and Genjutsu: they transform a source video, not text-to-video.
How to choose among the seven
Cheapest with sound: Wan 3.0 and Omni Flash tie at $1.25. Omni always has audio and caps at 10 seconds, so a 10-second clip is its maximum. Wan has no such cap, takes references and can go to 30 seconds. Seedance Mini is the cheapest token-priced option at $1.89, Seedance 2.5 the most expensive at $5.78 but the only one that reaches 30 seconds with Seedance's reference set.
Verify at run time
Rates and ranges can change, so read supported_durations, supported_resolutions and supported_aspect_ratios from GET /v1/videos/models before a batch. sume/auto also defaults to 720p and 8 seconds, with 3 to 10 second clips at 16:9 or 9:16, so a 10-second vertical clip is inside its range.
Sources
Related posts
More in Models
- AI image model news, October 10 2026: what changed for Sume users
Read on 2026-10-10: fal now lists Nano Banana 2.1 at $0.040 for 1K, Imagen is shut down in the Gemini API, and Hy Image 3.5 Preview is on OpenRouter, not Sume.
- Does Seedance 2 Fast do 1080p on Sume? No: 480p and 720p only
Seedance 2 Fast and Mini no longer list 1080p on Sume. Only seedance-2 and seedance-2.5 do. The exact refusal text and where to move a 1080p job.
- The 5 aspect ratios all 19 Sume image rows accept (a portable set)
All 19 Sume image rows list 1:1, 16:9, 9:16, 4:3 and 3:4. Use those five for a model bake-off and every row will accept the request without a 400.
- Gradium's 1,421 voices: how you pick a voice on Sume TTS
Gradium listed 1,421 TTS voices on Oct 7. Sume does not carry Gradium; here is how voice selection works on Sume TTS with avatar handles or voice ids.
Written by Sume