- Seedance 2.5 is ByteDance’s newest AI video model. It turns a text description, photos, video clips or audio into a video clip with sound — up to 30 seconds in one go.
- You can feed it up to 50 references at once (up to 30 images, 10 videos and 10 audio clips), so a character or product looks the same from shot to shot.
- Where to use it: ByteDance’s Dreamina site for subscribers in Europe, Asia, the Middle East and South America; Jimeng AI and Doubao Pro in China. It was not listed for the United States at launch.
- Cost: Dreamina puts it at about $0.097 per second of 720p video on its annual Advanced plan — roughly $2.91 for a full 30-second clip.
ByteDance, the company behind TikTok and CapCut, has released Seedance 2.5, the follow-up to Seedance 2.0, whose name alone is searched tens of thousands of times a month. The headline change is length: where most AI video tools still stop at five to fifteen seconds, Seedance 2.5 makes a 30-second clip in a single generation.
This guide explains, in plain language, what Seedance 2.5 does, where you can use it, what it costs, and what to watch out for. Everything below comes from ByteDance’s own Dreamina pages and from launch reporting; we have not run our own tests.
What is Seedance 2.5?
Seedance is an AI video generator: you describe a scene, or show it pictures and clips, and it creates a new video. Version 2.5 launched on 31 July 2026, according to TechNode’s launch report and ByteDance’s Dreamina announcement.
Three things set it apart from the tools most people have tried:
- Longer clips. Up to 30 seconds in one run. Longer videos no longer have to be stitched together from short pieces that don’t quite match.
- Lots of references. You can give it up to 50 “references” in a single request — photos of a person, a product shot, a video showing the camera movement you want, a voice recording. ByteDance’s split is up to 30 images, 10 videos and 10 audio clips.
- Editing instead of starting over. Dreamina says you can edit “specific timestamps, characters or other elements within a video”. If one detail is wrong — the colour of a jacket, a moment in the middle — you fix that part instead of throwing away the whole clip.
Dreamina also lists Seedance 2.5 as generating video with audio, so dialogue, music and sound effects come out with the picture rather than being added afterwards.
Where can you use Seedance 2.5?
Availability depends on where you live, and this is the part that confuses most people.
- Dreamina (dreamina.capcut.com) is ByteDance’s international creative site. Its launch post says Seedance 2.5 is rolling out to subscribers aged 16 and over in Europe, Asia, the Middle East and South America. The same post names the United States only for ByteDance’s image model, not for Seedance 2.5.
- Jimeng AI and the Pro version of Doubao carry it in China, according to TechNode.
- Developers can expect it on ByteDance’s Volcano Engine platform; TechNode reported the API as “expected”, not live, at launch.
If you are in the US and the model does not appear, that is the rollout, not your account. Be careful with third-party sites that claim to offer “Seedance 2.5 free”: some are genuine resellers of ByteDance’s API, many are not, and you will be handing them your photos.
How much does Seedance 2.5 cost?
ByteDance prices it per second of finished video. On its Seedance 2.5 pricing page, Dreamina quotes $0.097 per second on the annual Advanced plan, which works out to about $2.91 for a 30-second clip. That figure is for 720p video with a reference clip, the baseline Dreamina uses for its comparisons.
At launch Dreamina ran a promotion of four free Seedance 2.5 generations; that offer has ended. Free trials and credit bundles change often, so check the plan page before paying. For how this compares with Veo, Kling and others, see our AI video generation pricing breakdown.
What about 4K?
Dreamina’s product page advertises Seedance 2.5 as a “4K” generator. Its own price comparison, though, is quoted at 720p, and that is the resolution most API resellers list as the model’s native output. The safe reading: 4K is the top setting on offer, not what every generation gives you by default. If sharpness matters for your project, check the resolution option before you spend credits.
How to get good results
You don’t need special commands. A few habits help with any AI video tool, and they matter more with a model that accepts this many inputs:
- Start with the story in one or two sentences. Who is in the scene, where, and what happens from start to finish.
- Use references for anything that must stay the same — a face, a logo, a product. Several clear photos from different angles beat one blurry one.
- Show camera movement with a clip rather than describing it in words, if you have one.
- Fix, don’t regenerate. Use the edit tools on the part that is wrong. It is cheaper and keeps the parts you liked.
- Try a short test first. At about ten cents a second, a 30-second clip is not free; check the look on something shorter.
Seedance 2.5 vs other AI video tools
| Model | Longest clip (as announced) | Sound | How most people use it |
|---|---|---|---|
| Seedance 2.5 (ByteDance) | 30 seconds | Yes | Dreamina, Jimeng, Doubao |
| MiniMax H3 (Hailuo 3.0) | 15 seconds, up to 2K | Yes, stereo | Hailuo AI app and website |
| Veo 3.1 (Google) | Short clips | Yes | Gemini app, Flow |
| Kling (Kuaishou) | Short clips | Yes | Kling app and website |
For a fuller comparison, including which ones have a genuinely usable free tier, see the best AI video generators and the best free AI video generators.
If you have been using OpenAI’s Sora: OpenAI is retiring the Sora 2 models from its API on 24 September 2026. Our Sora alternatives guide covers where to move.
Things to be careful about
- Real people. Uploading someone else’s face or voice to make them “appear” in a video can be illegal and is against Dreamina’s rules. Use your own likeness, or people who have agreed.
- Labels. ByteDance says Seedance videos carry invisible watermarks and C2PA content credentials that mark them as AI-generated. Platforms such as TikTok and YouTube may label them automatically.
- Your uploads. Anything you give as a reference is sent to ByteDance’s servers. Read the privacy terms before uploading photos of children or private documents.
- Age limit. Dreamina’s rollout is for users aged 16 and over.
Frequently asked questions
Is Seedance 2.5 free?
Not in general. A launch promotion gave four free generations, and it has ended. Dreamina prices Seedance 2.5 per second of video; on the annual Advanced plan it quotes about $0.097 a second. Check Dreamina’s pricing page for current trials.
Can I use Seedance 2.5 in the United States?
Not at launch through Dreamina: ByteDance’s rollout post lists Europe, Asia, the Middle East and South America for the video model. That may change, so check Dreamina directly.
How long can a Seedance 2.5 video be?
Up to 30 seconds in a single generation, according to ByteDance and TechNode. That is about double what most rival tools produce in one go.
Does Seedance 2.5 make sound?
Yes. Dreamina describes it as generating video with audio, so speech, music and effects come out together with the picture.
Is Seedance 2.5 the same as CapCut?
No. CapCut is ByteDance’s video-editing app and Dreamina is its AI creation site. Seedance is the AI model that powers video generation inside ByteDance’s tools.