Seedance 2.5 is out: what shipped, and where you can actually use it
Updated July 31, 2026
What ByteDance shipped
Seedance 2.5 keeps 2.0's unified audio-video generation architecture and pushes on three fronts the Seed team names explicitly: long-form narrative, multimodal reference and editing. The framing in their own announcement is that users stopped wanting “a clip” and started wanting “a finished piece”.
| Seedance 2.0 | Seedance 2.5 (confirmed) | |
|---|---|---|
| Single generation | 4–15s per segment | 30s, with multi-round extension into minutes |
| Reference inputs | 9 images, 3 videos, 3 audio | 30 images, 10 videos, 10 audio in one request |
| Editing | Prompt-driven, whole-clip | Timestamp-addressed; green screen, viewpoint, camera move, reference |
| Reference types | Subject, style, motion | Adds white-model (greybox) layout with physical lighting derived from it |
| Look | Native audio, 480p–4K | Systematic work on the “greasy” AI look; fewer stray subtitles and stray BGM |
The three capabilities, concretely
1. Thirty seconds, then minutes
A single generation now runs 30 secondsand organises several logically connected shots inside it — setup, development, turn, resolution — rather than extending one image. Multi-round extension then continues from a finished clip while holding the subject, setting and pacing, so several minutes of material share one audiovisual language. ByteDance's demo of this is a continuous take that follows a singer from her dressing room, through the backstage corridor, onto the stage and out to a full arena.
2. Fifty reference assets in one shot
30 images, 10 videos and 10 audio clipscan travel with a single request — enough to cast an entire scene. Their concert example references a hall, a pianist, a cellist, a violinist, a lead singer, six orchestra sections, four choir groups and four audience blocks, all in one prompt, and keeps each subject's appearance and voice stable in the same frame.
3. Editing addressed by timestamp
You specify what happens in which second at generation time, and after generation you can point at a segment and change the character, the action, the sound or the plot beat while continuity holds. Around that sit green-screen editing (swap the world behind a preserved subject, with cloth, hair, gait and light re-rendered for the new physics), viewpoint and camera-move editing (keep the performance, redesign the camera), and reference editing.
Where you can use it — checked on launch day
- Jimeng AI (即梦) — web, video generation, select Seedance 2.5. Rolling out.
- Doubao Pro (豆包专业版) — video generation, select Seedance 2.5.
- Volcano Ark API — announced as coming “soon”. We checked the ModelArk model list on launch day: it carries doubao-seedance-1.0-pro through 2.0-mini and no 2.5 entry.
- Internationally — nothing. Dreamina's generator still exposes only 2.0-generation models, and neither fal.ai nor kie.ai lists a 2.5 endpoint.
What it looks like
Two clips ByteDance published with the launch. The first is the headline capability in one file — thirty seconds, one take, no cuts, with sound. The second is the short film the Seed team says Seedance 2.5 generated on its own, and it is the more revealing of the two: four and a half minutes of dialogue-driven scenes built by extension, which is exactly the workflow 2.5 was designed for.
The headline capability, unedited: a continuous 30-second generation with sound, holding a single performer against fire and shifting light without a cut.
Built by multi-round extension into four and a half minutes with dialogue, subtitles and a consistent cast — published at 360p by ByteDance, so judge the storytelling rather than the resolution.
Official Seedance 2.5 demos · ByteDance Seed · source · mirrored and re-encoded here, sound intact.
Official prompts from the launch
ByteDance published its demo prompts, which tells you more about how 2.5 wants to be driven than any spec sheet. Two patterns stand out.
Timestamps carry the direction. The Peking-opera demo is written as a schedule — “0–5s: opening close-up, camera orbits and transitions to a mid shot… 6–10s: stable orbit around the water sleeves… 11–20s: the martial role enters with a backflip” — with each beat naming which reference image supplies which character.
References are addressed like variables. Prompts point at @图片 1 (image 1), @视频 1 (video 1) and @白模 1 (white model 1) by number, then say what each one governs — this one supplies the camera path and blocking, that one supplies material, lighting and colour. The full set of launch prompts, translated and annotated, is in our Seedance 2.5 prompt library.
Where it is being used already
Two industry threads came with the launch. In education, Doubao Classroom is already generating lesson video — turning historical context, scientific principles and experiment procedures into moving demonstrations. In industry, the pitch is synthetic training data: robotics perception, industrial simulation and autonomous-driving scenarios, including rare events like extreme weather and unusual road conditions that are hard to collect in the real world.
Seedance 2.5 or MiniMax H3?
Both launched on July 31, and the contrast is unusually clean:
| Seedance 2.5 | MiniMax H3 | |
|---|---|---|
| Max single generation | 30s, extendable to minutes | 15s |
| Reference budget | 30 images, 10 videos, 10 audio | 9 images, 3 videos, 3 audio |
| Resolution | Up to 4K (2.0 lineage) | Native 2K |
| API today | No — “soon” | Yes, day one |
| Price | Not published | From $0.14/s at 2K (Atlas Cloud; fal $0.26) |
| Open weights | No | Promised within days |
Longer and more referenceable versus buyable and cheap. Full breakdown on the MiniMax H3 tracker, and all three together in Seedance 2.5 vs H3 vs FLUX 3.
Get the API opening, not the noise
One email when the Volcano Ark API opens or an international endpoint appears, with our measured cost per clip attached. We check the ModelArk catalog, fal and kie every day.
Frequently asked questions
- Is Seedance 2.5 out yet?
- Yes. ByteDance's Seed team released it on July 31, 2026 with an official project page at seed.bytedance.com/seedance2_5. It is rolling out inside Jimeng AI (web: video generation → select Seedance 2.5) and Doubao Pro. The Volcano Ark API is described as coming 'soon' and, as of our check on launch day, no doubao-seedance-2-5 model ID exists in the ModelArk catalog yet.
- What is confirmed in Seedance 2.5?
- Three things, from ByteDance directly: 30-second single generations with multi-round extension into several coherent minutes; multimodal reference of up to 30 images, 10 videos and 10 audio clips in one request; and timestamp-precise editing — you specify what happens at which second, then re-edit a character, action, sound or plot beat inside a finished clip without breaking continuity. It also adds green-screen editing, viewpoint and camera-move editing, and white-model (greybox) reference with physically-derived lighting.
- Can I use Seedance 2.5 outside China?
- Not yet. On launch day the only named channels are Jimeng AI and Doubao Pro, both Chinese products. Dreamina's generator still offers only 2.0-generation models, and neither fal.ai nor kie.ai lists a 2.5 endpoint. ByteDance paused the global rollout of Seedance 2.0 in March 2026 after cease-and-desist letters from Disney, Paramount, Warner Bros. and the MPA, so an international delay is a known pattern, not a surprise.
- How much does Seedance 2.5 cost?
- No pricing has been published. It is bundled into Jimeng and Doubao Pro rather than sold per second, and the Volcano Ark API rate card will only appear when the API does. We publish measured cost-per-clip within hours of the API opening — the same way we did for every other model in our cost calculator.
- What were the July 6 and July 16 dates people reported?
- Wrong. Chinese media reported, on insider sourcing, that Seedance 2.5 would reach the Jimeng experience centre on July 6, open to large accounts on July 13 and open its API on July 16. None of those was ever announced by ByteDance and none happened — we checked and published that on July 31, hours before the real launch. The actual release date was July 31, 2026, and the API still is not open.
- Seedance 2.5 or MiniMax H3?
- They launched the same day with opposite trade-offs. Seedance 2.5 goes further on length and reference volume — 30 seconds single-shot, extendable, with 50 reference assets against H3's 15 seconds and 9 images + 3 videos + 3 audio. H3 you can buy today from $0.14/s at 2K; Seedance 2.5 has no API and no price. If you need it in a pipeline this week, that decides it.