Ijharul Islam
Seedance 2.0 is ByteDance's AI video model that generates video and sound together from text, images, video clips and audio, making clips of 4 to 15 seconds. It was released in China in early February 2026 and has since rolled out in CapCut, Dreamina and ByteDance's API, and in third-party tools such as QuickVeed.
This guide covers what Seedance 2.0 is, what makes it different from Veo and Kling, where you can use it, the rules that come with it, and how to prompt it, with a worked example.
The short answer
Seedance 2.0 is a video generation model from ByteDance Seed, the research team behind TikTok's parent company. It accepts four kinds of input (text, images, video and audio) and generates video with synchronized sound in clips of 4 to 15 seconds, natively at 480p or 720p. You can use it in CapCut, Dreamina, ByteDance's BytePlus API and tools such as QuickVeed.
Key takeaways
- Seedance 2.0 comes from ByteDance Seed and was released in China in early February 2026.
- Its standout feature is mixed references: up to 9 images, 3 video clips and 3 audio clips in one generation.
- Clips run 4 to 15 seconds with sound, natively at 480p or 720p.
- In CapCut it carries a visible watermark and C2PA Content Credentials, and early rollouts blocked uploads with real faces.
- QuickVeed offers Seedance 2.0, 2.0 Fast and 1.5 Pro, all with native audio.
Disclosure: we make QuickVeed and show how to do this in it, along with other ways. Details about other products come from their public pages, checked in October 2026 and linked at the end.
What is Seedance 2.0?
Seedance is ByteDance's family of video generation models, made by its Seed research team. Version 2.0 is described by ByteDance as a native multi-modal audio-video generation model, which means two things. It makes picture and sound in the same pass, rather than adding audio afterwards. And it can take four kinds of input at once: text, images, video and audio.
That second point is what makes Seedance 2.0 different. Most video models take a prompt and maybe a first frame. Seedance can also take a reference clip for the motion or camera move, a few images for the look of a character or product, and an audio clip for the rhythm or voice, and combine them in one generation. ByteDance's paper lists the limits on its open platform as up to 3 video clips, 9 images and 3 audio clips.
| Spec | Seedance 2.0 (per ByteDance) |
|---|---|
| Maker | ByteDance Seed |
| Released | Early February 2026 (China); wider rollout from March 2026 |
| Inputs | Text, image, video, audio |
| References per generation | Up to 9 images, 3 video clips, 3 audio clips |
| Clip length | 4 to 15 seconds |
| Native resolution | 480p and 720p |
| Sound | Generated with the video, in sync |
| Aspect ratios | Six, according to CapCut's announcement |
Tools that offer Seedance usually also list Seedance 2.0 Fast, a quicker and cheaper variant, and the earlier Seedance 1.5 Pro.
Where you can use Seedance 2.0
| Where | Notes |
|---|---|
| CapCut (AI Video, Video Studio) | For paid CapCut users. Rolled out first in Indonesia, the Philippines, Thailand, Vietnam, Malaysia, Brazil and Mexico, then more markets in March 2026, Japan on March 31 and the US on April 7, 2026 |
| Dreamina and Pippit | ByteDance's creative tools, named in CapCut's announcement |
| BytePlus API | For developers; ByteDance's page links a playground and API access |
| Third-party tools | Several video tools, including QuickVeed, offer Seedance models next to Veo and Kling |
Availability still differs by country and plan, so check inside the app you plan to use. In China the model is offered through ByteDance's own apps.
Which route to pick? If you already edit in CapCut and are on a paid plan in a supported country, it is the most direct way in, and you get the full set of reference inputs. If you are a developer, the BytePlus API gives you programmatic access. If you want to compare Seedance with Veo, Kling and Runway on the same prompt, a multi-model tool is simpler than holding four subscriptions.
Rules and safeguards to know
CapCut's April 2026 announcement lists the safeguards that come with Seedance 2.0 in its apps:
- Real faces: in the initial rollout, you can't generate from images or videos that contain real faces.
- Copyrighted characters and IP: the system is designed to block unauthorized generation of protected IP.
- Watermarks: generated content carries a visible watermark, an invisible watermark and C2PA Content Credentials.
- Content rules: nudity, graphic content and harmful misinformation are blocked, with checks before and after generation.
For business use, the practical effects are: plan on brand characters, products and places rather than real people; don't try to recreate famous characters; and if your video shows realistic events that didn't happen, label it as AI-generated where platforms ask you to.
How to prompt Seedance 2.0
Because Seedance takes mixed references, think of the prompt as directions that tell the model what each reference is for. A good structure:
- Say what each reference does. “Use image 1 for the product, image 2 for the kitchen setting, video 1 for the camera move.”
- Describe the action. One clear action per shot, with a speed: “slowly slides the jar across the counter.”
- Describe the camera and light if no reference video sets them.
- Describe the sound, in its own sentence: dialogue with the speaker named, then effects and ambience.
- Set the length to fit the action: 4-6 seconds for one move, 10-15 for a short scene.
Worked example: a 10-second product clip
| Input | What you give |
|---|---|
| Image 1 | Clean photo of a honey jar on a white background |
| Image 2 | Photo of a sunlit wooden kitchen counter |
| Prompt | Place the honey jar from image 1 on the counter from image 2. A hand lifts the lid and a wooden dipper draws out a slow ribbon of honey. Close-up, slow push in, warm morning light. Soft kitchen ambience and a gentle clink of the lid. |
| Length and shape | 10 seconds, 9:16 for Reels and TikTok |
If the label on the jar warps, shorten the clip, reduce the motion, or add “keep the jar and label unchanged”. Text on packaging is the hardest thing for any model to keep stable.
Good first projects for Seedance 2.0
Seedance's strengths are long clips with sound and the way it follows references. These projects play to them:
| Project | Setup | Why Seedance fits |
|---|---|---|
| A 10-15 s product scene for Reels | Product photo plus a setting photo, motion and sound in the prompt | Long enough for a small story without a cut |
| A matching set of clips for one campaign | The same product and style images in every generation | Image references keep the look consistent |
| Recreating a camera move you like | A reference clip of the move plus your product image | Video references guide motion and framing |
| A short scene timed to a beat | An audio reference plus images | Audio references can shape rhythm and timing |
| Quick social B-roll | Text prompt only, 4-6 s, on Seedance 2.0 Fast | Cheaper and quicker for simple shots |
Three habits help. First, label every reference in the prompt so the model knows what it is for. Second, keep references consistent with each other: the same light, the same colour palette, the same time of day. Third, keep the action simple even when the clip is long. A 15-second clip with one slow, clear action looks better than one with five.
And because CapCut's version blocks real faces and adds a visible watermark, plan campaign work around products, places and stylised or fictional characters.
How to use Seedance 2.0 in QuickVeed
QuickVeed offers Seedance 2.0, Seedance 2.0 Fast and Seedance 1.5 Pro, all with native audio, next to the other models in its generator.
- Open the AI video generator (in the app it is under Create video).
- Type a prompt, or upload an image to use as the starting frame.
- Choose a model. Clip lengths run from 2 to 15 seconds depending on the model: Veo 3.1 gives 4, 6 or 8 s, Kling v3 3 to 15 s, Seedance 2.0 4 to 15 s, Runway Gen-4.5 5 or 10 s.
- Pick the aspect ratio (16:9, 9:16, 1:1, 4:3, 3:4 or 21:9 where the model supports it) and the resolution (480p to 1080p).
- Check the price shown, then generate. Failed generations are refunded, and every result is saved in your asset library.
Pick Seedance 2.0 in step 3 for 4 to 15 seconds with sound, or Seedance 2.0 Fast to draft more cheaply. The exact price shows before you generate, and failed generations are refunded. For comparison, here are some fixed reference prices:
| Example clip in QuickVeed | Credits |
|---|---|
| Vidu Q3 Turbo, 2 s, 540p | 20 |
| Luma Ray 3.2, 5 s, 540p | 38 |
| Kling v3, 5 s, 720p | 210 |
| Veo 3.1 Fast, 8 s with audio | 300 |
| Veo 3.1, 8 s with audio | 800 |
In QuickVeed you start from a prompt or one image. If you need Seedance's full mixed-reference input (several images plus video and audio clips), use a tool that exposes all of it, such as CapCut or the API.
Seedance 2.0 vs Veo 3.1 vs Kling 3.0
| Seedance 2.0 | Veo 3.1 | Kling 3.0 | |
|---|---|---|---|
| Maker | ByteDance | Google DeepMind | Kuaishou |
| Length per clip | 4-15 s | 4, 6 or 8 s | 3-15 s |
| Native resolution | 480p, 720p | 720p, 1080p, 2160p | 720p, 1080p |
| Sound | Yes | Yes | Yes |
| Inputs | Text, image, video, audio | Text, image, video (extend) | Text, image, video for Elements |
| Stands out for | Mixing many references | Polished short shots with dialogue | Multi-shot and longer scenes |
None is best at everything. Use Seedance when you have reference material that should shape the result, Veo for short premium shots, and Kling for 10-15 second scenes. More detail: what Kling AI is and what Google Veo 3 is.
Frequently asked questions
Who makes Seedance 2.0?
ByteDance, the company behind TikTok and CapCut. It was built by ByteDance's Seed research team.
Is Seedance 2.0 free?
Not generally. In CapCut it is for paid users, and the API is paid. Some tools include it in a trial; QuickVeed's 14-day trial has 75 credits, no card needed.
How long can Seedance 2.0 videos be?
4 to 15 seconds per generation. Join clips in an editor for longer videos.
Is Seedance 2.0 available in the US?
CapCut's announcement says it reached paid CapCut users in the US on April 7, 2026. It is also available through third-party tools and the BytePlus API.
Does Seedance 2.0 generate audio?
Yes. It generates video and sound together, in sync, and can also take audio clips as references.
Try it with your own video
Try Seedance 2.0 with one product photo and a short motion prompt. Upload a product photo, pick Seedance 2.0, and judge the motion and sound for yourself. QuickVeed plans start at $18/month, and new accounts get a 14-day trial with 75 credits, no card needed. Sign up or see pricing.
Sources
Competitor details in this post were checked against these public pages in October 2026. Products and prices change, so confirm with each vendor before you decide.
