Longest single shot per request: Vidu, Veo, Grok, Wan, Seedance
How long one request can run: Vidu Q4 16 s, Veo 8 s, Grok 15 s, Wan 3.0 and Seedance 2.5 30 s. Sume's matching row for each and the longest-shot cost.

The longest single shot per request is 30 seconds, on Wan 3.0 (Alibaba and Sume) and Seedance 2.5 (Sume). Vidu Q4 stops at 16, Grok Imagine at 15, HappyHorse 1.0 at 15, Gemini Omni Flash 1.1 at 10 and Veo 3.1 at 8.
The table gives each vendor's limit as read on 2026-10-10 and the Sume row that matches, if any.
The limits side by side
Vendor limits come from the pages listed in the sources. Sume limits come from the Video Generation docs and the capability table in the API code. Where a vendor lists a range, the table shows the maximum.
| Model | Vendor maximum | Sume row | Sume maximum |
|---|---|---|---|
| Vidu Q4 | 16 s | None | None |
| Veo 3.1 | 8 s | veo-3.1 (if listed) | 8 s |
| Grok Imagine video | 15 s | grok-imagine-video-1.5 | 15 s |
| Wan 3.0 | 30 s | wan-3.0 | 30 s |
| HappyHorse 1.0 | 15 s | None | None |
| Seedance 2.5 | Not fetched | seedance-2.5 | 30 s |
| Gemini Omni Flash 1.1 | Not fetched | gemini-omni-flash-1.1 | 10 s |
Why 30 seconds is not always the answer
A long single shot is useful when you want continuity, such as a walk-through or a slow reveal. But it is also the most expensive way to fail. A 30-second Seedance 2.5 clip at 720p bills $17.334 on Sume, and a 30-second Wan 3.0 clip at 720p bills $3.75. If the first five seconds are wrong, you have paid for all 30.
Test with a short clip on the same row, then scale the duration. The Sume rates are linear in seconds for Wan, so the test costs a fixed fraction of the final shot.
| Sume id | Resolution | Longest shot | Billed cost |
|---|---|---|---|
| wan-3.0 | 480p | 30 s | $1.875 |
| wan-3.0 | 720p | 30 s | $3.75 |
| wan-3.0 | 1080p | 30 s | $7.50 |
| seedance-2.5 | 480p | 30 s | $8.06031 |
| seedance-2.5 | 720p | 30 s | $17.334 |
The Omni exception
Sume's docs line that each other catalog model tops out at 15 seconds is not true for Gemini Omni Flash 1.1, which accepts 3 to 10 seconds. If you plan a 12-second shot, Omni will refuse it. Pick Seedance 2.5, Wan 3.0, MiniMax H3 (5 to 15 seconds) or the Grok row (up to 15 seconds).
Read supported_durations in GET /v1/videos/models at run time instead of hard-coding the number, since catalogs change.
Choosing for a duration
Under 10 seconds with 4K or a reference clip: Omni. Up to 15 seconds from a still: Grok on Sume is the cheapest. Over 15 seconds: only Wan 3.0 and Seedance 2.5 qualify on Sume, and Wan is much cheaper per second at every resolution. Vidu Q4's 16 seconds would give one more second than Grok, but Vidu is not on Sume. The Video Generation page lists every limit.
Splitting a long shot instead
If a 30-second shot is too costly to risk, you can split it into two 15-second requests on a row that supports it and join them in an editor. Each half costs half as much on a per-second basis, and you can reject a bad half without paying for the other. The cut will show unless you plan it, so choose a moment such as a camera whip or an object passing the lens.
On Sume, Wan 3.0 is the practical choice for this approach at $0.125 per second at 720p, so two 15-second halves bill $1.875 each.
Sources
- Vidu Q4 AI Video Model with Native Audio (read 2026-10-10)
- Generate videos with Veo 3.1 in Gemini API (read 2026-10-10)
- Video Generation: Grok API Capabilities (read 2026-10-10)
- How to use Wanx wan3.0 video generation model (read 2026-10-10)
- HappyHorse text-to-video API reference (read 2026-10-10)
- Video Generation
Related posts
More in Comparisons
- Luma Ray 2 5-second keyframes: Sume rows with start and end frames
Luma's API makes Ray 2 clips up to 5 s from start and end keyframes. Sume has no Luma row; these Sume rows accept both frames and cover a 5-second clip.
- MAI-Voice-2.1 clones from seconds of audio: what Sume's API takes
Microsoft says MAI-Voice-2.1 clones a voice from seconds of audio. Sume does not carry it, and voice cloning on Sume is app-only; the API takes voice ids.
- MAI-Voice-2.1 pricing vs Sume TTS: what Sume lists, what it does not
MAI-Voice-2.1 costs $22 per 1M characters per Microsoft AI. Sume does not list MAI; its TTS runs Sonic voices at $47.50 per 1M characters.
- Meta skips AI labels: no significant edit, no real-looking human
Meta applies no AI label when its own tools make no significant edit and no photorealistic human. See how Sume trim, crop and dim relate, from Meta's text.
Written by Sume