Castaly · Models · Video
Seedance 2.5: Specs, Pricing and Examples
Generate video with Seedance 2.5 in Castaly: up to 1080p and 30 seconds, built-in audio, 30 image references, video editing and extension, with a live price.
Models & Pricing
Prices are in USD. The listed group rate applies before any request-specific multiplier. Your account may have different access or discounts.
| Group | Price | Conditions | Unit | Price |
|---|---|---|---|---|
| Media | Output | 1080p | Per 1M tokens | USD 14.4495 |
| Media | Output | 1080p_input_video | Per 1M tokens | USD 8.645000000000001 |
| Media | Output | 480p | Per 1M tokens | USD 13.2145 |
| Media | Output | 480p_input_video | Per 1M tokens | USD 7.904000000000001 |
| Media | Output | 720p | Per 1M tokens | USD 13.2145 |
| Media | Output | 720p_input_video | Per 1M tokens | USD 7.904000000000001 |
Seedance 2.5 is ByteDance's video model, and in Castaly it offers the longest clips and the most references of the four Seedance models. It renders up to 1080p and 30 seconds, takes up to 30 reference images, 10 videos and 10 audio clips, and can edit or extend a video you already have. The live price is on this page.
Specifications
| Setting | Options |
|---|---|
| Model ID | seedance-2.5 |
| Resolution | 480p, 720p, 1080p |
| Length | Auto, or any whole number of seconds from 4 to 30 |
| Ratio | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive |
| Start from | A prompt, a first or last frame, or a set of references |
| References | Up to 30 images, 10 videos and 10 audio clips |
| Workflows | Auto, reference, edit, extend |
| Audio | Generate audio switch, no extra cost |
| Output | MP4 or MOV, with the option to return the last frame as an image |
| Billing | Per token |
Sample
The prompt behind this clip:
"A small research submersible descends through the deep ocean, its headlights sweeping over a drifting field of bioluminescent jellyfish and a sunken stone archway. Camera follows from behind, particles float in the beams, blue-black water, calm and mysterious, photoreal, no text."
What you can do with it
- Generate from references. Combine images, videos and audio clips in one request, and name them in the prompt, for example a character from an image performing a movement from a video.
- Edit a video. Give it a source clip and an instruction, such as removing something from the background.
- Extend a video. Continue a clip you already have.
- Guide with frames. Set a first frame, a last frame, or both. Frames and references cannot be mixed in one request.
- Chain shots. Ask for the last frame of the result as an image and use it as the first frame of the next shot.
Start from auto and the model works out the task from your media, or pick a workflow yourself.
How it is billed
Seedance 2.5 bills by tokens, not seconds. The token count comes from the canvas and the length of the clip, and nothing in the prompt changes it. A reference video adds its own length to the count, at a lower input-video rate. Auto duration holds enough for the longest possible clip and returns the unused part when the clip settles.
Seedance 2.5 pricing works through what a clip costs at each resolution. Failed tasks are refunded in full.
Seedance 2.5 or another model?
- Seedance 2.0 if you need 4K, or a lower rate and fixed lengths up to 15 seconds are enough.
- Kling 3.0 for per-second billing and 4K with a simple motion-control input.
- Wan 3.0 if you also need clips up to 30 seconds but want per-second billing, or want to start from a file or a web page.
The Seedance comparison covers all four Seedance models side by side.
FAQ
How long can a Seedance 2.5 video be?
Up to 30 seconds. Choose any whole number of seconds from 4 to 30, or use Auto duration and let the model decide.
Does Seedance 2.5 render 4K?
No. It stops at 1080p. Seedance 2.0 is the Seedance model that renders 4K.
Does audio cost extra on Seedance 2.5?
No. The token rate is the same whether Generate audio is on or off.
Can I use Seedance 2.5 through an API?
Yes. The same model ID works through the Modelflare API with the same account and balance. The request format is in the API documentation.