Seedance 2.5 'extend twice': what Sume's 30-second request does
ByteDance says Seedance 2.5 makes up to 30 seconds with two extensions. Sume's seedance-2.5 id takes 4-30 seconds per request and has no extend field.

ByteDance's Seedance 2.5 page says you can 'create videos up to 30 seconds in a single generation, with the option to extend twice'. On Sume, seedance-2.5 takes 4-30 seconds in one request, but the catalog has no extend field and no way to hand a finished clip back to be extended. To go past 30 seconds you generate a second clip and join them.
The vendor wording is from Seedance 2.5 on ByteDance Seed, read on 2026-10-03. The page does not say how long each extension is, so this post does not either. Sume's range comes from the Video generation docs and the Video Router docs.
What maps and what does not
The single-generation half maps directly. Everything beyond it does not, and the table shows the difference in one place.
Sume's seed and unknown-field handling is strict: a field the model does not list, such as an extend or extensions key, is rejected with a 400 rather than ignored, so you find out at request time instead of after a bill.
| Capability | ByteDance page | Sume seedance-2.5 |
|---|---|---|
| Single generation length | up to 30 s | 4-30 s |
| Extend a finished clip | extend twice | no extend field |
| Resolutions | not stated on the page | 480p, 720p, 1080p |
| Start and end frames | not stated on the page | first_frame and last_frame |
| Reference video, image, audio | reference video understanding described | supported within the 12 item combined cap |
Joining a second clip
The practical route to a longer sequence is a chain. Generate clip one, extract its last frame, and send that frame as the first_frame of clip two with a prompt that continues the action. The video-frames tool returns stills from a finished clip, and the timeline tool joins the clips into one file.
Expect a small visual seam. The model does not see the previous clip's motion or audio, only a still, so keep the camera move and lighting words identical in both prompts.
Cost of a chain
Each clip is billed on its own: list price times 1.25, by the model's own unit. Two 30 second generations cost what two 30 second generations cost; there is no discount for continuing. If the story is 45 seconds, generating 30 plus 15 is cheaper than 30 plus 30 and trimming, because you pay for seconds generated.
For a quick decision: if a story fits inside 30 seconds, ask for it in one request and skip the seam entirely. If it does not, plan the cut points before you generate, not after.
Sources
Related posts
More in Models
- Seedance 2.5 weak spots: physics and crowded scenes, a test plan
ByteDance says Seedance 2.5 still struggles with complex physics and multi-subject scenes. A 480p test plan with frame checks you can run on Sume.
- Reference video or motion transfer: Seedance 2.5 or Genjutsu?
ByteDance says Seedance 2.5 reads a reference video beyond motion transfer. When to send reference_video_urls to Seedance and when to use Genjutsu on Sume.
- Seedance 2.5 timestamp-level editing: what Sume exposes today
ByteDance describes timestamp-level editing and 30-second clips for Seedance 2.5. See which of those fields Sume's Video Router accepts for seedance-2.5.
- Seedance 2.5 video-to-video: a reference clip is not a person swap
Seedance 2.5 on Sume takes reference videos to condition a new 4 to 30 second generation. To swap people in an existing clip, send it to h3-max-recast instead.
Written by Sume