Choose a category

3 models available
8 models available

3 models available
5 models available
6 models available
3 models available

8 models available








Seedance 2.0
ByteDance's flagship multimodal video model, up to 4K.
Start / End Frames
Optional. Animate from a first frame, or add a last frame to land on. Use frames or references (images, video, audio), not both. JPEG, PNG or WebP, up to 30MB each.
Creative Vision
Reference uploads as @Image1, @Image2 … in the prompt.
Video Parameters
AI audio synced to the video. Does not change the price.
Cinema Preview
Generation History0 records
Seedance 2.0 AI Video Generator
Seedance 2.0 is ByteDance's February 2026 video model, and it works differently from the rest of this site: you upload reference images and then point at them inside the prompt as @Image1, @Image2, casting them as your character or your location. Four to fifteen seconds, 480p up to 4K, audio included.
Upload the references first, then write the prompt around them — @Image1 only means something once an image is in slot one
Why Choose Seedance 2.0
Released by ByteDance's Seed team on 10 February 2026, Seedance 2.0 was built as a unified multimodal system rather than a text-to-video model with an image option bolted on. Reference images are addressable objects in the prompt, which is a different way of directing a shot.
Cast Your Shot With @Image
Upload up to five references and address them individually: "@Image1 and @Image2 for the character, @Image3 for the alley." Roles are assigned explicitly instead of being guessed from a single blended reference.
480p Through to 4K
Four resolution tiers on one page. The 480p tier exists to be thrown away — it is the draft — while 4K is there for the one take that has to hold up on a large screen.
Audio at No Extra Cost
Sound is generated with the picture and does not change the price, so there is no reason to turn it off. Unusual on this site: the Kling models charge double for the same switch.
Six Aspect Ratios
16:9, 9:16, 1:1, 4:3, 3:4 and 21:9. The 21:9 tier is worth knowing about — with the 4K setting it is the closest thing on this site to a deliverable cinema frame.
A 20,000-Character Prompt
Four times the room of any other model here. That is not padding: a five-reference, multi-beat prompt with per-shot camera direction genuinely runs long, and this is the only page that will take it.
Failed Runs Are Refunded
Credits are held at submission and returned automatically if generation fails or the provider never calls back. That matters most here, where a single 4K run is the largest charge on the site.
Draft at 480p Before You Commit
Twenty-five credits at 480p and five seconds tests your references and your casting. The same clip at 4K is 235. Get the prompt right on the cheap tier first — that price gap is the whole workflow.
How to Use Seedance 2.0
The reference workflow is what makes this model worth the price, and it is also what people skip. A prompt with no @Image tags turns Seedance into an ordinary and expensive text-to-video model.
Upload Your References
Up to five images, 30MB each. Order matters, because it determines the numbers: the first upload is @Image1. A clean character shot and a clean location shot beat five noisy ones.
Assign Roles in the Prompt
State what each reference is for before describing the action: "@Image1 for the woman, @Image2 for the courtyard. She walks in from the left as the camera tracks with her." Unassigned references get used loosely.
Test at 480p
Five seconds at 480p is 25 credits. It is too soft to judge detail and perfectly adequate to judge whether the model cast your references correctly — which is the thing that actually goes wrong.
Render the Final Take
Move to 1080p at 115 credits for five seconds, or 4K at 235, once the casting and the motion are right. Rendering takes several minutes at the top tiers; you can leave the page while it works.
Rehearse on Mini, Perform on Standard
Seedance 2.0 Mini takes identical prompts and references for 5 credits at five seconds. Prompt engineering there and rendering here is the cheapest route to a good 4K clip.
Seedance 2.0
FAQ
Reference images, the @Image syntax, resolutions up to 4K, audio and credit costs for the Seedance 2.0 model.
Seedance 2.0 is a video generation model released by ByteDance's Seed research team on 10 February 2026. It is a unified multimodal system: text and reference images are combined in a single generation pass, with role-based asset tagging so a specific reference can be assigned to a specific element of the shot. This page runs the standard tier, which is the one that reaches 4K. FlowVeo3 is an independent platform and is not affiliated with ByteDance.
Upload images, then refer to them by position in the prompt text. The first upload is @Image1, the second @Image2, and so on up to @Image5. Write something like "@Image1 for the character's face and outfit, @Image2 for the workshop interior" and the model treats each as the source for that element. Without the tags it still uses the images, but it decides for itself what they are for.
No — references are optional. With no images uploaded, Seedance 2.0 behaves as a straightforward text-to-video model. But that is an expensive way to use it: if you are not going to reference anything, Wan 2.6 or Grok Imagine will give you a text-driven clip for a fraction of the credits.
From 20 to 705 credits, priced per second and rising steeply with resolution. At five seconds: 25 credits at 480p, 50 at 720p, 115 at 1080p and 235 at 4K. Fifteen seconds at 4K is 705 credits, the largest single charge available on this site. The exact figure for your settings is shown before you submit.
Only for delivery. The 4K tier costs roughly nine times the 480p tier at the same length, and it buys you resolution, not better casting, motion or prompt adherence. Everything you are actually iterating on is visible at 480p. Treat 4K as the render step for a shot you have already approved.
Same controls, same prompt syntax, same reference handling — the difference is the resolution ceiling and the price. Standard reaches 4K. Fast and Mini both stop at 720p, and Mini is roughly a fifth of Standard's price at the same settings. Use Mini or Fast to develop the prompt, then run it here for the final resolution.
Yes. Under Advanced you can add up to 3 reference videos and up to 3 reference audio clips, each type limited to 15 seconds in total. Refer to them in your prompt as @Video1, @Audio1 and so on. Reference videos are billed by length at a lower per-second rate; reference audio is free. Use them together with reference images, not with start/end frames. Only upload media you have the rights to use.
Several minutes, and noticeably longer at 4K and fifteen seconds. You can leave the page while it renders and the clip appears in your history when it is ready. Finished videos stay available for seven days, so download anything you plan to keep — particularly at these prices.
Try Seedance 2.0
Upload your references, cast them with @Image tags, and test the whole idea for 25 credits at 480p.