AI Video Generator
Imagifly — Seedance 2.0 / 2.5
Seedance is ByteDance Seed's video generation line. The 1.0 technical report landed in June 2025, the family moved to 2.0 (with Fast and Mini variants) and reached 2.5 on 31 July 2026. No weights are published: models are reachable only through Volcengine Ark in China or BytePlus ModelArk internationally.
- Text to Video
- Image to Video
- ByteDance Seed
From 210 credits per video

About Seedance 2.0 / 2.5
Seedance is ByteDance Seed's video generation line. The 1.0 technical report landed in June 2025, the family moved to 2.0 (with Fast and Mini variants) and reached 2.5 on 31 July 2026. No weights are published: models are reachable only through Volcengine Ark in China or BytePlus ModelArk internationally.
Seedance 1.0 Pro handles text-to-video and image-to-video under one model id, at 480P or 1080P, in 5- or 10-second clips at 24 FPS, with 16:9, 9:16, 4:3, 3:4, 1:1 and 21:9 frames and native multi-shot cuts. The technical report measures a 5-second 1080P clip at 41.4 seconds on an NVIDIA L20.
Version 2.0 extended single-pass generation to 4-15 seconds and added first/last-frame control, all-modal references and video editing or extension. Version 2.5 reaches 4-30 seconds with multi-round extension, generates native audio and video together, supports 11 prompt and dialogue languages, accepts up to 50 reference assets (30 images, 10 videos, 10 audio clips) and offers a 480p draft mode before finishing at 1080p.
What it is good at
- One model for text-to-video and image-to-video, with first/last-frame control
- 2.5 generates up to 30 seconds in one pass and can be extended
- 2.5 produces native audio and video together in 11 dialogue languages
- Up to 4K output, and up to 50 reference assets on 2.5
Technical specs
- Developer
- ByteDance Seed
- Released
- 2025-06-10
- Latest version
- Seedance 2.5 (2026-07-31)
- License
- Proprietary, API-only
- Architecture
- Not disclosed by ByteDance (1.0 has a technical report, arXiv:2506.09113)
- Parameters
- Not disclosed (proprietary, API only)
- Task
- Text-to-video and image-to-video (native audio from 2.5)
- Recommended settings
- 1.0 Pro: 480P or 1080P, 5 or 10 seconds, 24 FPS. 2.0: 4-15 seconds, 480p to 4k, with Fast and Mini limited to 480p/720p. 2.5: 4-30 seconds, 480p to 4k (10-bit) with native audio. Exact frame rates and tiers follow each API's documentation.