Seedance 2.5 Reference-to-Video

Generate a single take of up to 30 seconds from a text prompt plus up to 30 reference images and audio files. Up to 1080p.

By ByteDance

Pricing: $1.3968 per second

Overview

Seedance 2.5 Reference-to-Video by ByteDance takes a prompt and up to 30 reference images, then holds that cast, set and palette across a take of up to 30 seconds. Each reference is addressed in the prompt as @Image1, @Image2 and so on.

Key capabilities

• **From $0.26 per second** of generated video at 480p, $0.57 at 720p and $1.40 at 1080p. You pay for the seconds you get back.
• **Multi-region failover**, Runflow runs optimized inference across multiple regions. If one region has an outage, traffic routes to another automatically.
• **Up to 30 reference images**, addressed as @Image1 to @Image30. JPG, PNG, WebP, BMP, TIFF, GIF, HEIC and HEIF, up to 30 MB each.
• **Reference audio**: attach up to 10 audio files as @Audio1 onward to drive a voice, a song or a rhythm.
• **Takes up to 30 seconds** with lip-synced native audio.
• **Seven aspect ratios**: 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, or auto.

Seedance family

| Variant | Input | Use it for |

|---|---|---|

| Seedance 2.0 Text-to-Video | text prompt | multi-shot clips up to 4K from a brief |

| Seedance 2.0 Reference-to-Video | text + up to 9 images | keeping a product or character consistent |

| Seedance 2.5 Text-to-Video | text prompt | single takes up to 30 seconds |

| Seedance 2.5 Reference-to-Video | text + up to 30 images | long takes with a locked cast and set |

Technical specifications

| Spec | Details |

|---|---|

| References | up to 30 images, plus up to 10 audio files, 50 files in total |

| Resolutions | 480p, 720p, 1080p |

| Duration | 4 to 30 seconds, or auto |

| Audio | native, with lip-sync |

| Frame rate | 24 fps |

| Output format | MP4 (H.264) |

| Latency | about 3 to 7 minutes for a 5 second 720p clip |

Examples

• Two reference images, a character and a mountain landscape, held through a wide establishing shot that pushes in to a close-up

Frequently asked questions

What is Seedance 2.5 Reference-to-Video?

Seedance 2.5 Reference-to-Video is a ByteDance video model that generates a take of up to 30 seconds from a text prompt and up to 30 reference images. The references fix how each subject looks for the whole clip.

How do I point the prompt at my reference images?

Order your image URLs in the image_urls array, then call them @Image1, @Image2 and so on inside the prompt. Reference audio works the same way with @Audio1.

How much does Seedance 2.5 Reference-to-Video cost on Runflow?

You pay per second of generated video: $0.2646 at 480p, $0.5676 at 720p and $1.3968 at 1080p. A 5 second 720p clip costs about $2.84. References add no extra cost.

How long does a Seedance 2.5 Reference-to-Video run take?

A 5 second 720p clip takes about 3 to 7 minutes. More references and longer takes add time. Submit the run and poll the run ID until it reports succeeded.

Can I use Seedance 2.5 Reference-to-Video output commercially?

Yes. Output from Seedance 2.5 Reference-to-Video is licensed for commercial use. You keep the rights to the videos you generate.

Do I need to manage GPUs to run Seedance 2.5 Reference-to-Video?

No. Runflow runs Seedance 2.5 Reference-to-Video for you. You send one API call with your image URLs and get a video URL back.

How do I get started with Seedance 2.5 Reference-to-Video?

Create a Runflow API key, then POST your prompt and image_urls to /v1/models/bytedance/seedance-2.5/reference-to-video/runs and poll the returned run ID.

Related models

• Seedance 2.5 Text-to-Video, Generate a single coherent take of up to 30 seconds with native audio from a text prompt. Up to 1080p, seven aspect ratios.
• Seedance 2.0 Reference-to-Video, Generate video with native audio from a text prompt plus up to 9 reference images. Up to 4K, 4 to 15 seconds, seven aspect ratios.
• Happy Horse Reference-to-Video, Generate 1080p video with synchronized native audio from a text prompt and references. Aspect ratios: 16:9, 9:16, 1:1, 4:3, 3:4. Duration: 3–15s.
• Happy Horse Image-to-Video, Alibaba's #1-ranked Happy Horse 1.0 — generate 1080p video with synchronized native audio and multilingual lip-sync from text prompts or images.

Discoverable surfaces

• Dispatch endpoint: POST https://api.runflow.io/v1/models/seedance-2.5/reference-to-video/runs
• Per-model spec (markdown): https://app.runflow.io/models/seedance-2.5/reference-to-video/llms.txt
• Docs page: https://docs.runflow.io/models/seedance-2.5/reference-to-video
• Public OpenAPI spec: https://docs.runflow.io/api/openapi.public.json
• Agent skill (start here): https://www.runflow.io/.well-known/agent-skills/runflow/SKILL.md