ByteDance Seedance video models
The Seedance family offers frame-based generation and multimodal routes that combine images, video, and audio. Your choice depends on whether the shot needs simple start and end frames, a collection of references, or a longer story. Compare the supported input modes before choosing an output resolution.
5 models available in Starpop. Updated .
Compare ByteDance video models
| Model / best use | Inputs | Output options | Estimated cost |
|---|---|---|---|
| ByteDancevideoSeedance 1.5 Pro Designed transitions between two frames | Text, Image | 720p, 1080p4–12s | $0.075–$0.15 USD / second0.75–1.5 credits / second |
| ByteDancevideoSeedance 2 Multimodal campaign hero shots | Text, Image, Video, Audio | 480p, 720p, 1080p4–15s | $0.15–$0.65 USD / second1.5–6.5 credits / second |
| ByteDancevideoSeedance 2.5 Native-audio stories up to 30 seconds | Text, Image, Video, Audio | 480p, 720p4–30s | $0.25–$0.50 USD / second2.5–5 credits / second |
| ByteDancevideoSeedance 2 Fast Rapid multimodal iteration | Text, Image, Video, Audio | 480p, 720p4–15s | $0.10–$0.25 USD / second1–2.5 credits / second |
| ByteDancevideoSeedance 2 Mini Low-cost multimodal storyboards | Text, Image, Video, Audio | 480p, 720p4–15s | $0.075–$0.15 USD / second0.75–1.5 credits / second |
Choosing a ByteDance model
Starting from designed frames
Seedance 1.5 Pro is a focused option for native-audio shots with optional start and end frames. Use a clear transition between those frames instead of asking the prompt to replace their composition.
Comparing multimodal drafts
Seedance 2 Mini and Fast suit reference testing at their supported resolutions. Keep each reference's purpose consistent across variations, then consider standard Seedance 2 when the deliverable needs its higher-resolution option.
Building a longer sequence
Consider Seedance 2.5 when the brief needs a longer clip or a larger reference pack. Write events chronologically and approve a shorter test before spending credits on the complete sequence.