NOW LIVESeedance 2.5 is now live on I2V.aiTry it now
i2v.ai
  • Studio
  • Models
  • Seedance 2.5
  • Pricing
Models / Video

ByteDance

Seedance 2.5

seedance2-5

Text, image, keyframe, and multimodal reference video generation

Text to videoImage to videoFirst and last frame480p / 720pAudio output
Run with API
PlaygroundPricingAPIExamplesOverviewFAQ

Detailed descriptions of scenes, actions, and styles will result in better generation quality

4 seconds5 seconds30 seconds

Loading...

Join Discord community
Content policy notice
Each provider applies its own content policy. Some models may reject unsafe requests or certain real-person reference inputs. If a task is rejected, review the model guidance and try permitted source material.

Price by mode and output

Credits are charged from the same Studio and API wallet. USD estimates below use $9.90 Mini Pack; larger packs or promotions can lower your effective cash cost.

ModeConfigurationCreditsEstimated USD
Text or image input480p · 5s · without audio54$0.69
Text or image input480p · 5s · with audio54$0.69
Video input example480p · 5s output · 1 × 5s input · without audio33$0.42
Video input example480p · 5s output · 1 × 5s input · with audio33$0.42
Text or image input720p · 5s · without audio116$1.47
Text or image input720p · 5s · with audio116$1.47
Video input example720p · 5s output · 1 × 5s input · without audio70$0.89
Video input example720p · 5s output · 1 × 5s input · with audio70$0.89
Modes
Text to video, Image to video, First and last frame, Reference images, Multimodal reference, Video to video
Duration
4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 seconds
Resolution
480p, 720p
Aspect ratios
16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive
Audio
Supported

Build with Seedance 2.5

Create a task, keep the returned task ID, then poll recordInfo until the result is ready. Exact model pricing is included so you can estimate product cost before you run.

Get an API key
POST /api/v1/jobs/createTask
{
  "model": "seedance2-5",
  "inputs": {
    "prompt": "A cinematic tracking shot through a futuristic city at sunset",
    "duration": "5s",
    "aspectRatio": "16:9",
    "resolution": "480p"
  }
}

Two ways to direct a scene

30-second narrative continuity

See the model hold onto camera movement, materials, light, and character motion. Open each example for the full prompt.

Reference generationImages × 8
Reference 1 used for the narrative continuity example.
1/8
Output

Multi-reference image to video · continuous camera move

Eight references become one uninterrupted walk through six connected rooms, while the character and architectural rhythm stay consistent.

1 / 6

Prompt:One continuous take. The camera steadily follows a person in a black coat (reference @image1) walking from left to right through six connected rooms, each with a different colour palette and atmosphere. Every room shares the same structure: white walls, pale herringbone wood flooring, French double-height windows, and sheer white curtains, referencing @image2. The exterior view and interior mood change completely in every room. The subject walks at a constant speed through every open doorway. 0-5s: the first room is an American-comic fight. The subject enters and fights a character from @image3; the character is defeated. 5-10s: the second room feels warm and is made of felt. Outside is a sunflower field from @image4. Warm orange light fills the room while a painter paints sunflowers from @image5. The subject becomes felt as they enter. 10-15s: the third room is sad, rendered as black-and-white comic stop motion. Rain falls outside; the light is cool, grey, and low. A person sits alone on the floor, head down and knees hugged, beside a phone showing an unanswered call. When the subject enters, the lights go out, then immediately come back on in colour as flowers instantly grow across the room. 15-20s: the fourth room is joyful and submerged in the sea, referencing @image6. As the subject swims in, they become transparent; coral reefs and schools of fish surround them. 20-25s: the fifth room is surprising. Outside is a fireworks-filled night sky from @image7, and coloured flashes reflect through the room. A crowd surrounds the subject and pulls them into the celebration. 25-30s: the subject reaches a blank final room, stands in the centre, and snaps their fingers with a matching snap sound. The image cuts to black and the word "seedance" appears in the centre, referencing @image8. Cinematic visual language, premium fashion-film treatment. Let the exterior view determine the light in each room for strong emotional contrast. No other on-screen text.

Expand

Full prompt

One continuous take. The camera steadily follows a person in a black coat (reference @image1) walking from left to right through six connected rooms, each with a different colour palette and atmosphere. Every room shares the same structure: white walls, pale herringbone wood flooring, French double-height windows, and sheer white curtains, referencing @image2. The exterior view and interior mood change completely in every room. The subject walks at a constant speed through every open doorway. 0-5s: the first room is an American-comic fight. The subject enters and fights a character from @image3; the character is defeated. 5-10s: the second room feels warm and is made of felt. Outside is a sunflower field from @image4. Warm orange light fills the room while a painter paints sunflowers from @image5. The subject becomes felt as they enter. 10-15s: the third room is sad, rendered as black-and-white comic stop motion. Rain falls outside; the light is cool, grey, and low. A person sits alone on the floor, head down and knees hugged, beside a phone showing an unanswered call. When the subject enters, the lights go out, then immediately come back on in colour as flowers instantly grow across the room. 15-20s: the fourth room is joyful and submerged in the sea, referencing @image6. As the subject swims in, they become transparent; coral reefs and schools of fish surround them. 20-25s: the fifth room is surprising. Outside is a fireworks-filled night sky from @image7, and coloured flashes reflect through the room. A crowd surrounds the subject and pulls them into the celebration. 25-30s: the subject reaches a blank final room, stands in the centre, and snaps their fingers with a matching snap sound. The image cuts to black and the word "seedance" appears in the centre, referencing @image8. Cinematic visual language, premium fashion-film treatment. Let the exterior view determine the light in each room for strong emotional contrast. No other on-screen text.

Why it works

The prompt fixes the repeated structure and walking rhythm first, then lets each room vary only where the narrative asks it to.

Output goal

A single, coherent 30-second camera move that preserves one character across changing worlds.

Tips

  • Use the same subject reference across every scene beat.
  • Describe the invariants before describing each scene change.
  • Keep the camera path explicit when asking for a continuous shot.

What is Seedance 2.5?

Seedance 2.5 is a video model from ByteDance. In practical terms, the Seedance family is designed for controllable AI video generation from text, images, and supported reference media. Text, image, keyframe, and multimodal reference video generation.

On I2V, Seedance 2.5 supports Text to video, Image to video, First and last frame, Reference images, Multimodal reference, Video to video. Available output options include 480p, 720p, with aspect-ratio behavior shown as 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive. The same public model ID and credit balance work in Studio and through the task API.

What Seedance 2.5 can do

Input workflows

Seedance 2.5 exposes Text to video, Image to video, First and last frame, Reference images, Multimodal reference, Video to video. The model-locked Studio form only shows controls that apply to this model.

Output control

Configured output resolution: 480p, 720p. Supported aspect-ratio behavior: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive.

Reference-led creation

Use source images, frames, or reference media when the selected mode accepts visual input.

Best uses for Seedance 2.5

Short-form storytelling

Create self-contained shots for social posts, trailers, mood films, and narrative prototypes.

Product and campaign motion

Explore product reveals, camera movement, environments, and campaign concepts before a full production shoot.

Storyboard previsualization

Turn prompts or approved frames into moving references that help teams evaluate pacing and shot direction.

When to choose Seedance 2.5

Choose this model when shot direction and input flexibility matter. Compare Standard, Fast, Mini, and Pro variants by supported modes, duration, resolution, and credit cost. I2V shows the exact valid configurations before you run the task, so model selection can be based on both capability and cost.

Seedance 2.5 FAQ

What is Seedance 2.5?+

Seedance 2.5 is a ByteDance video model available in I2V Studio and through the I2V task API. It supports Text to video, Image to video, First and last frame, Reference images, Multimodal reference, Video to video.

What inputs and outputs does Seedance 2.5 support?+

Seedance 2.5 supports Text to video, Image to video, First and last frame, Reference images, Multimodal reference, Video to video. Output resolution is 480p, 720p, with aspect-ratio behavior listed as 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive. The live form is restricted to valid options for this model.

How much does Seedance 2.5 cost on I2V?+

Seedance 2.5 starts at 33 credits ($0.42 using the current Mini Pack reference) for Video input example, 480p · 5s output · 1 × 5s input · without audio. The pricing table lists other valid configurations.

Can I use Seedance 2.5 through an API?+

Yes. Use the public model ID seedance2-5 with POST /api/v1/jobs/createTask, then poll GET /api/v1/jobs/recordInfo until the task succeeds or fails. Studio and API share one credit balance.

Keep exploring

Related models

View all models

Seedance 2.0 Mini

Released: June 2026

Text, image, and multi-reference video generation with audio

Audio on · 480p · 5s

$0.23

Video input · 480p · 5s

$0.28

Audio on · 720p · 5s

$0.48

Video input · 720p · 5s

$0.58

Open model

Seedance 2.0 Fast

Released: April 2026

Text, image, reference, and video-to-video generation

Standard · 480p · 5s

$0.37

Video input · 480p · 5s

$0.43

Standard · 720p · 5s

$0.77

Video input · 720p · 5s

$0.91

Open model

Seedance 2.0

Released: February 2026

Text, image, reference, and video-to-video generation

Standard · 480p · 5s

$0.46

Video input · 480p · 5s

$0.56

Standard · 720p · 5s

$0.96

Video input · 720p · 5s

$1.18

Open model
Resources
  • Blog
  • Studio
  • Scenes
  • Works
  • Prompts
  • Image to Prompt
  • Batch Image to Prompt
Company & Legal
  • About
  • Contact
  • Privacy Policy
  • Terms of Service
  • Refund Policy
Image Models
  • Qwen Image 2
  • Seedream 4.5
  • Seedream 5.0
  • Nano Banana Pro
  • Nano Banana Flash
  • Nano Banana 2
Video Models
  • Google Veo 3.1
  • Google Veo 3.1 Lite
  • Google Veo 3.1 Pro
  • Seedance 1.5 Pro
  • Seedance Fast
  • Seedance Quality
  • Seedance 2.0
  • Kling v3.0
  • Kling v3.0 Pro
i2v.ai

I2V AI — image and video creation, with API access

DiscordEmail

This website is an independent third-party platform. We are not affiliated with, endorsed by, or officially representing any AI model providers referenced on this site. All trademarks and brand names belong to their respective owners.

© 2026 I2V AI All Rights Reserved.

Privacy Policyprivacy@i2v.aisupport@i2v.ai