Dreamina Seedance is the AI video model ByteDance sells through Dreamina, the creative app in the CapCut family. Seedance 2.0 arrived in February 2026 and Seedance 2.5 followed on July 31, 2026 with 30-second clips and up to 50 reference files. We have run Seedance through its API while building our own video tools, so this guide mixes the public specs with what we saw in our own tests. EAST.IM is not affiliated with Dreamina, CapCut or ByteDance.
What Dreamina Seedance is
Seedance is a family of video models made by ByteDance. You give it a text prompt, a starting image or a set of reference files and it returns a short video with sound. Dreamina is the consumer app where most people use it. Developers can also reach the same models through API platforms such as BytePlus, fal, Replicate and Cloudflare.
The name causes some confusion. "Dreamina Seedance 2.5" and "Seedance 2.5" are the same model. Dreamina simply puts its own name in front because it is the official app for it.
Seedance 2.5 vs Seedance 2.0
| Seedance 2.0 | Seedance 2.5 | |
|---|---|---|
| Released | February 2026 | July 31, 2026 |
| Clip length | about 4 to 15 seconds | up to 30 seconds in one generation |
| Reference files | up to 12 | up to 50 (images, video, audio and text) |
| Sound | yes | yes, with fewer random subtitles and stray music according to Dreamina |
| Edit part of a clip | video extension and basic edits | change one moment or one character without redoing the whole video |
| Versions in Dreamina | Pro, Fast and Mini | one model |
The longer clip is the change people notice first. Before 2.5, a 30-second video usually meant stitching two or three shorter generations together and hoping the face and lighting matched. Now one pass can cover it.
Resolution is where the numbers stop lining up. Dreamina markets Seedance 2.5 as native 4K and also mentions a 180-second long video mode in beta. The API versions we checked, including Cloudflare's model page, only offer 480p and 720p for 2.5. The app and the API are not the same product, so a spec you read on one may not apply to the other.
Who can use Seedance 2.5
At launch, Dreamina rolled out Seedance 2.5 to paid members aged 16 and over in Europe, Asia, the Middle East, Africa and South America. The United States was left off that first list, which is why many guides written in early August say it is not available there.
Dreamina later announced on its X account that Seedance 2.5 is live for US users, and its own pricing page now lists the US among the countries where the Seedance offers apply. If you are in the US and still cannot see it, the model may simply not be on your account yet. Dreamina says eligibility is not guaranteed even in listed countries.
Outside Dreamina, businesses can reach Seedance 2.5 through API resellers. BytePlus does not serve the US, but other platforms do.
Is Seedance free?
Partly. Dreamina gives out a small number of free credits, enough to try the app. Seedance 2.5 launched for paid members, so most people will need a plan to use it regularly.
Dreamina sells monthly and yearly plans that come with credits, and every generation spends credits based on the model, length and resolution. Seedance 2.5 costs more per second than Seedance 2.0 Fast or Mini. Prices change by country and Dreamina runs frequent first-month discounts, so check Dreamina's Seedance price page for the current numbers in your region. Read the renewal price, not only the first month.
Dreamina also adds a visible AI label, an invisible watermark and C2PA Content Credentials to the videos it makes.
Why Seedance rejects photos of real people
This is the limit most people run into first. Seedance 2.0 and 2.5 do not accept reference images or videos that show a real person's face.
The rule goes back to February 2026. Within days of the Seedance 2.0 launch, Disney sent ByteDance a cease and desist letter over videos of its characters, and other studios followed. ByteDance responded by pausing uploads of real people, adding filters for well-known characters and labeling every output.
We hit the block ourselves. When we sent Seedance 2.5 a normal photo of a person through the API, it refused before generating anything and returned this error:
InputImageSensitiveContentDetected.PrivacyInformation
In Dreamina the message is worded differently, usually a note that the image may contain a real person. Either way, sending the same photo again does not help. This is how the model is meant to work, not a bug.
What still works:
- Text prompts that describe a person instead of showing one
- Illustrated, animated or clearly stylized characters
- Objects, animals, places and products
What we saw when we tested it
Our test used Seedance 2.0 Fast through the API with a stylized character portrait and a short clip as references. The goal was to see whether it could put that character into the clip and follow the original movement.
It took 99 seconds to return a 4-second video at 864 x 496. The character came through clearly, with the face and makeup intact. The shot itself did not. Seedance turned the scene into a half-body portrait against a plain gray studio wall, gave the character a coat and props from the reference image and kept the mouth open the whole time. The source clip started with a closed mouth and opened slowly. None of that timing carried over.
That matches what Seedance is built for. It is very good at taking references and making a new shot from them. It is not built to copy an existing performance frame by frame. If you want a new scene with a consistent character, it is a strong pick. If you want to put someone into a scene that already exists, it is the wrong tool.
Seedance vs Kling, from our own use
We have run both models on the same kind of job, so here is the short version.
Seedance reads a lot of references at once and builds a fresh shot. Composition, background and pacing are its own choices. It works best when you are making something new.
Kling is better when you already have a clip and want the result to follow it. Its motion control and video edit modes keep the original framing, timing and camera movement, and it accepts photos of real people.
Neither one is better across the board. They solve different problems.
If you want yourself in a trend video
Most people searching for Seedance want one thing: their own face in a viral clip. Seedance will not take your photo, so that route is closed.
Our AI video templates are built for exactly this. Each one is a single trend, and you upload a photo of yourself to star in it. A few popular ones:
- Gas station dance for three people
- Burning Bridges AI, the Drake party scene
- Zombie hug, the last hug story for two people
- Alors on danse AI for two people
Videos need a paid plan, and failed generations are refunded automatically. Only upload photos of yourself or of people who have agreed to it.
FAQ
What is Dreamina Seedance?
It is ByteDance's AI video model, offered in Dreamina, the creative app from the CapCut family. It turns text, images, video and audio references into short videos with sound.
Is Seedance 2.5 available in the US?
Yes. It was not part of the July launch, but Dreamina later announced US availability and its pricing page lists the US. It can still take time to show up on every account.
What is the difference between Seedance 2.0 and 2.5?
Seedance 2.5 makes clips up to 30 seconds instead of about 15 and takes up to 50 reference files instead of 12. It can also edit one moment or one character in a clip without redoing the whole video.
Is Seedance free to use?
Dreamina gives a few free credits to try it, but Seedance 2.5 launched for paid members. Plans and credit prices vary by country.
Why does Seedance say my photo contains a real person?
ByteDance blocks reference images and videos that show real human faces. The rule started after Hollywood studios complained in February 2026, and resubmitting the same photo will not change the result.
Can I put my own face in a Seedance video?
Not through a normal upload. If you want to star in a trend video, use a tool that accepts your own photo, such as our AI video templates.