Seedance 2.5 on BytePlus excludes the US: check access first
BytePlus lists Seedance 2.5 on ModelArk in supported markets excluding the United States. What that means for a US team, and how to read the Sume catalog.

No, not directly. The BytePlus Seedance 2.5 page says ModelArk availability covers supported markets globally, excluding the United States. A US team should not assume it can sign up for the direct API, and should check each other access path separately.
This post is about reading availability claims, not about legal advice. The facts below come from the vendor pages read on 2026-10-03.
What the BytePlus page says
The BytePlus page is the primary source for what Seedance 2.5 is. Availability is one line among several capability lines, so it is easy to miss when you are comparing clip length and reference limits.
| Item | What the page states |
|---|---|
| Clip length | Up to 30 s |
| References | Up to 50 multimodal references per clip |
| Editing | Timestamp-level and character-level secondary editing |
| Safety | C2PA Content Credentials, visible watermarks, real-face restrictions |
| ModelArk availability | Supported markets globally, excluding the United States |
Third-party listings are separate questions
Other platforms also list the model. The Runway changelog for Oct 2, 2026 describes Seedance 2.5 Draft Mode in Agent, and the Higgsfield API post lists Seedance 2.5 at $0.0738 per second.
A platform listing is not the same as availability for your account. Each platform has its own terms, and a model listed on one page can be gated by region, plan, or review on another. Ask each vendor, in writing, before you plan a launch around a single path.
What the Sume catalog tells you
Sume documents seedance-2.5 in its video catalog with 4 to 30 second clips at 480p, 720p, and 1080p. The catalog tells you what the API can call. It does not tell you what your legal or procurement team will approve, so keep those two questions apart.
import os
import requests
resp = requests.get(
"https://api.sume.com/v1/videos/models",
headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
timeout=30,
)
resp.raise_for_status()
for model in resp.json()["data"]:
if model["id"].startswith("seedance"):
print(model["id"], model["supported_durations"], model["supported_resolutions"])Questions to put to any access provider
Whatever route you pick, the same handful of questions applies. They are cheap to ask before a launch and expensive to answer after one, and a written answer from the provider is worth more than a screenshot of a pricing page.
Keep in mind that the BytePlus page also states real-face restrictions and watermark rules. Those apply to the content you generate, regardless of where the account lives, so availability is only the first gate. The second is what your brief contains, and the third is what your delivery channel accepts.
- Is the model available to accounts in your country, and to the company entity that will be billed?
- Does the provider apply the same real-face and watermark rules as the original vendor?
- Which limits apply on this route: clip length, reference count, and resolution?
- What happens to a job that fails: is it billed, or refunded?
- Where do generated files live, and for how long?
A short decision list
Run the call above with your own key. If seedance-2.5 is in the response, the id is callable from your workspace and you can read its limits from the same row. The video generation docs describe every field in that response.
If your policy depends on where a model is served, record the question you asked and the answer you got next to the model id. A one-line note beats a guess six months later.
- Read the vendor availability line, not only the capability table.
- Check the access path you will actually use, not the one in the announcement.
- Confirm real-face and watermark rules for that path before you brief a campaign.
- Write the answer down with the date you got it.
Sources
Related posts
More in Models
- Self-host MiniMax H3 or use an API: a decision table
MiniMax released H3 as an open-weight omni-modal model on Jul 31, 2026. A decision table for running it yourself versus calling a hosted video API.
- Veo 3.1 extend adds 7 s up to 20 times, 720p only: plan a long clip
Google's Veo docs let you extend a clip by 7 seconds up to 20 times, at 720p only. Read the arithmetic, then compare it with one-call 30 second models on Sume.
- Voxtral Mini Transcribe 2 and Realtime v26.02: what Mistral lists
Mistral lists Voxtral Mini Transcribe 2, Voxtral Realtime v26.02 and Voxtral TTS v26.03. How to prepare video audio for any transcription model.
- Wan 3.0 doubles clip length to 30 seconds: Alibaba ids vs Sume
Alibaba Model Studio lists wan3.0-video and wan3.0-video-prime at 2 to 30 seconds, up from 15 on Wan 2.7. What Sume's wan-3.0 accepts and a 30 s request.
Written by Sume