30s Continuous Video
Create complete cinematic sequences without stitching clips together.
Native 4–30 second clips at 480P/720P, up to 50 multimodal references, timestamp-level editing and native audio — now in the Renoise Canvas, billed per generation from your plan credits.
Dreamina’s public Seedance 2.5 page frames the model as a controlled production workflow, not just a longer text-to-video generator.
Create complete cinematic sequences without stitching clips together.
Combine images, videos, audio, and style references in a single generation.
Edit specific elements while preserving the original scene, motion, and style.
Generate natural voices and synchronized lip movements across multiple languages.
Seedance 2.5 is framed around controllable production workflows: references, multimodal inputs, precision editing, stable 30-second generation and multilingual creation.
Use green-screen, white-model or motion references to guide character movement, spatial layout and interactions. The goal is stronger scene control than text prompts alone.
Create a native 30-second clip in one pass while preserving identity, lighting, camera movement and scene continuity across the full shot.
Combine prompts, scripts, reference photos, videos, music and style guides in one generation so the model can follow a complete creative direction.
Refine only the parts of a video that need adjustment while preserving lighting, composition, motion, audio and timeline continuity.
Create natively across 10+ languages including Chinese, English, Spanish, Arabic, Portuguese, Vietnamese, Japanese and Korean, with unified audio-video output at 480P/720P.
Dreamina positions Seedance 2.5 around publishable commercial and cinematic work: social content, product ads, brand campaigns and short-form storytelling.
Seedance 2.5 was officially released on July 31, 2026, going live first as the default video model in ByteDance’s own consumer apps, 即梦 (jimeng) and Dreamina. API pricing was published alongside it, with Model Playground and API access rolling out on 火山方舟 and BytePlus ModelArk in early August.
In Renoise, Seedance 2.5 landed in the Canvas model picker on August 7, 2026 and is billed per generation from your plan credits. Seedance 2.0 stays available for the jobs it still does better, notably native 4K.
June 23, 2026
ByteDance / Volcano Engine previewed Seedance 2.5 at the Force 原动力大会 and released the model’s first creative short film.
July 31, 2026
Seedance 2.5 officially launched, going live as the default video model in ByteDance’s 即梦 and Dreamina consumer apps.
Aug 7, 2026
The Seedance 2.5 entry ships in the Canvas model picker, alongside a one-week launch campaign (Unlimited Seedance 2.5 on Advanced) that closed on August 14, 2026.
Public Seedance 2.5 capabilities next to Seedance 2.0. Both run in the Renoise Canvas — they are different tools, not an upgrade path.
| Capability | Seedance 2.5 (Recommended) | Seedance 2.0 |
|---|---|---|
| Max clip length | 4–30s, 30s native in one shot | 4–15s |
| Asset inputs | Up to 50 combined (30 img + 10 video + 10 audio) | 9 images + 3 video + 3 audio |
| Audio-only reference | Supported on its own | Must be paired with another reference |
| Precision editing | Timestamp-level region editing | — |
| 3D blockout input | Yes | — |
| Resolution | 480P / 720P | Up to native 4K |
| Pick it when | You need length, references or dialogue | You need 4K or a fast short clip |
Generate with Seedance 2.5 in Renoise — native 4–30 second clips, R2V reference control and timestamp editing, billed per generation from your plan.
One Renoise plan unlocks Seedance 2.5 and 2.0, plus every other available video and image model.
Yes. ByteDance officially released Seedance 2.5 on July 31, 2026, first as the default video model in its own 即梦 and Dreamina apps, with the developer API rolling out in early August. In Renoise you can pick Seedance 2.5 in the Canvas model picker.
No. Unlimited Seedance 2.5 was a launch campaign that ran from August 7 to August 14, 2026: during that week, Advanced subscribers could generate through a separate Unlimited Seedance 2.5 entry in the model picker without spending credits per video. The window has closed and that entry is gone. Seedance 2.5 is now billed per generation from your plan credits, like every other model in the Canvas — there are no daily caps and no free tier.
The big changes are control and duration: 30-second standard videos, R2V references, up to 50 multimodal inputs (30 images + 10 video + 10 audio), timestamp-level region editing, native audio-video generation and native 10+ language creation.
R2V refers to reference-to-video control. Dreamina describes green-screen and white-model references that can guide motion, spatial layout and character interactions, so the generation follows a structured visual plan instead of relying only on text.
Yes. Seedance 2.5 is positioned around 30-second standard generation, which helps creators produce a full scene, ad beat or social cut with fewer stitched clips.
Up to 50 multimodal inputs in one generation. That can include prompts, scripts, reference photos, videos, audio, music and style guides, which is useful for brand work where the model needs to follow more than a single text prompt.
Dreamina highlights local video editing, where selected regions or objects can be refined without recreating the whole scene. It also describes white-model / blockout-style references for controlling motion and spatial composition.
No. Seedance 2.5 outputs at 480P or 720P — ByteDance’s official specs list no 4K or 1080P for it. If you need native 4K video, use Seedance 2.0, which supports 4K in Renoise today.
Seedance 2.5 is positioned for social media videos, product ads, ecommerce showcases, brand campaigns and cinematic short stories — workflows that benefit from longer scenes, stable characters, R2V motion guidance and local fixes.
Seedance 2.5 was officially released on July 31, 2026, launching first in ByteDance’s 即梦 and Dreamina apps. The developer API followed in early August on 火山方舟 and BytePlus ModelArk, and Seedance 2.5 is now an entry in the Renoise Canvas.
ByteDance has published Seedance 2.5 API pricing (BytePlus ModelArk, per million tokens): about $10.70 without a video input and $6.40 with one — roughly $0.51 for a 480p 5s clip and $1.16 for 720p 5s. In-app credit values in 即梦/Dreamina are not public. In Renoise you spend credits from a plan per generation, with no daily caps.
Yes — Seedance 2.5 is an entry in the Canvas model picker, billed per generation from your credits like any other model. Seedance 2.0 stays available alongside it, and it is still the one to pick when you need native 4K.
Seedance is developed by ByteDance, under Volcano Engine (豆包). It was unveiled at the Force conference on June 23, 2026, alongside its first creative short film.