Wan 3.0 edit and extension at Alibaba vs Sume's Wan row
Alibaba lists Wan 3.0 editing and extension; Sume's wan-3.0 row does text, image, end frame and references only. Edit with Omni Flash 1.1 or H3 Max Recast.
Alibaba's Wan 3.0 page lists video editing and video extension as modes (read 2026-10-10). Sume's wan-3.0 row does neither: its capability entry has video-to-video set to false, so for edits on Sume you use Gemini Omni Flash 1.1 through video_url or H3 Max Recast.
What Sume's Wan row does cover is text, image, end frame and multimodal references, up to 30 seconds.
What Alibaba lists
The Model Studio guide (read 2026-10-10) names two ids, wan3.0-video and wan3.0-video-prime. Modes are text-to-video, first and last frame, multimodal reference, video editing and video extension. Duration is 2 to 30 seconds plus -1 for smart duration, and resolution is 480P, 720P or 1080P with 1080P as the default. Audio is on by default, and the video URL is valid for 24 hours.
Reference limits are up to 10 images, 5 videos and 5 audio files, with up to 20 references overall.
| Mode | Alibaba | Sume wan-3.0 |
|---|---|---|
| Text-to-video | Yes | Yes |
| First and last frame | Yes | Yes |
| Multimodal references | Yes | Yes, 10 images, 5 videos, 5 audio |
| Video editing | Yes | No |
| Video extension | Yes | No |
| Prime variant | wan3.0-video-prime | Not listed |
Doing an edit on Sume
The Video Router docs describe an edit mode for gemini-omni-flash-1.1: send a prompt and a video_url. The constraint text says resolution is optional and defaults to 720p, aspect_ratio is rejected, and the duration is not sent; a duration you give only sizes the reservation. Edits bill output seconds like any Omni job, at $0.125 per second at 720p.
H3 Max Recast is a narrower edit: it swaps the main people in a source video, 5 to 30 seconds, for 1 to 4 photos, and keeps motion, camera and sound.
| Row | Edit type | Duration | Billed per second |
|---|---|---|---|
| gemini-omni-flash-1.1 | Prompt plus video_url | 3 to 10 s | $0.125 at 720p |
| h3-max-recast | Swap people | 5 to 30 s | 768p or 1080p, not priced here |
Extension without an extend mode
Sume has no extend call for Wan. The nearest equivalent is to generate the whole shot in one request, since the row accepts up to 30 seconds. If you must continue an existing clip, pass its last frame as a start image on a new request, or pass the clip as a reference video, which Wan accepts up to 15 seconds in total.
Neither recreates Alibaba's extend. The seam can show, and the style can drift, so preview the join.
- One long shot: wan-3.0, up to 30 s.
- Continue a clip: start frame from its last frame.
- Change an existing clip: Omni edit.
Cost check
A 30-second Wan 3.0 clip at 720p bills $3.75 on Sume, and a 10-second Omni edit at 720p bills $1.25. Check usage.cost, and read the Video Generation docs for the shared rules.
Why the Prime variant is absent
Alibaba lists a second id, wan3.0-video-prime, and Runway's pricing page (read 2026-10-10) separately lists wan3 and wan3_prime. Sume's registry has one Wan 3.0 id and no Prime entry. If your workflow names the Prime variant, you cannot select it through Sume, and you should not assume the Sume row is a stand-in for it. Test a clip and judge the result.
Sources
Related posts
More in Comparisons
- Wan 3.0 vs Seedance 2.5: the keep rate where Wan is cheaper
Wan 3.0 at 720p costs $0.63 per 5 s and Seedance 2.5 costs $2.89 on Sume. Wan wins per kept clip unless its keep rate falls below about 21.8% of Seedance's.
- Will Meta label a Sume video ad as AI? What Meta says it detects
Meta says it will detect third-party AI ads via industry-standard signals and add AI info to About this ad. The Sume docs promise nothing either way.
- Grok Imagine mid-video frame pin vs Sume's first and last frames
xAI lets Grok Imagine 1.5 pin first, last and mid frames. Sume's Grok row takes one start still; this maps pinning to Sume rows with first and end frames.
- YouTube's four disclosure triggers checked against Sume's tools
Four YouTube disclosure triggers: real person, altered real footage, invented realistic scene, music. Each is checked against Sume generation and media tools.
Written by Sume