Seedance 2.5: Longer Shots, Native Audio
The next generation of Seedance — clips up to 30 seconds with sound generated in the same pass. Launching on AutoDirector as soon as upstream access opens.
Seedance 2.5 is the next-generation model in ByteDance's Seedance family, and its headline change is length: where Seedance 2.0 tops out at 15 seconds, Seedance 2.5 generates clips up to 30 seconds in a single pass. That doubles what fits in one take — a full product story, a complete joke, an entire scene beat — without stitching multiple generations together.
The second major upgrade is native audio. Instead of generating silent footage and adding sound later, Seedance 2.5 produces audio in the same generation pass, keeping sound and picture in sync by construction.
AutoDirector has Seedance 2.5 wired into the platform already — the same three modes as Seedance 2.0 (text to video, image to video, reference to video) with launch resolutions of 480P and 720P. It goes live for everyone as soon as upstream API access opens; until then, every prompt and workflow you build on Seedance 2.0 will carry over directly.
Seedance 2.5 planned specs
| Generation modes | Text to video · Image to video · Reference to video |
|---|---|
| Duration | 4–30 seconds (any whole second) |
| Resolutions | 480P / 720P at launch |
| Aspect ratios | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 |
| Audio | Native audio generation |
| Status | Pre-integrated; awaiting upstream launch |
What changes from Seedance 2.0
Up to 30-second takes
Twice the maximum length of Seedance 2.0 — enough for a complete scene or product story in one uncut generation.
Native audio
Sound is generated with the picture in a single pass, so ambience and action stay in sync without post-production.
Same three modes
Text, image, and reference to video all carry over — your existing Seedance workflows apply on day one.
Familiar prompting
Prompts written for Seedance 2.0 transfer directly; longer durations mainly reward more explicit shot-by-shot structure.
Draft-friendly launch tiers
480P and 720P at launch keep iteration affordable; higher tiers arrive as the upstream API expands.
Day-one availability here
The integration is already built and tested behind a flag — when ByteDance opens access, it appears in the model list.
How to be ready for Seedance 2.5
- 1
Build your workflow on Seedance 2.0
The prompting style, modes, and parameters are the same family — everything you learn now transfers.
- 2
Structure prompts shot by shot
30-second generations reward explicit sequencing: describe each beat in order, with camera and lighting per shot.
- 3
Collect reference material
Reference to video carries over to 2.5 — a library of motion and character references pays off immediately.
- 4
Watch the model list
Seedance 2.5 will show up in the generator's model picker the moment it goes live — no separate signup needed.
Prompt ideas that use 30 seconds well
These are written for long single takes — try them on Seedance 2.0 in 15s form today.
One-take product story
Single continuous shot: a coffee bean falls in slow motion, camera follows it into a grinder, through swirling grounds, pulls back to reveal a finished latte on a sunlit cafe table, warm tones, smooth motion throughout
Scene with a turn
A violinist plays alone in an empty concert hall; lights slowly rise revealing a full orchestra joining in; camera cranes up to a wide shot as the music swells, cinematic, golden light
Long vertical hook
Vertical video: time-lapse of a street corner from dawn to night — vendors open, crowds build, neon flickers on, rain begins — one fixed camera position, rich color grading
Seedance 2.5 pricing
Pricing will be announced when Seedance 2.5 opens for public generation.
Until then, Seedance 2.0 offers the same modes today from 4 credits per second.
See current plansSeedance 2.5 FAQ
What is Seedance 2.5?
Seedance 2.5 is the next-generation AI video model in ByteDance's Seedance family. Its two headline upgrades over Seedance 2.0 are single-pass clips of up to 30 seconds and native audio generated together with the picture.
When can I use Seedance 2.5?
AutoDirector has the integration built and tested; it goes live as soon as ByteDance opens upstream API access. It will appear directly in the generator's model list — no separate waitlist.
How is Seedance 2.5 different from Seedance 2.0?
Twice the maximum duration (30s vs 15s) and native audio. Modes, aspect ratios, and prompting style stay the same, so existing Seedance workflows carry over.
What resolutions will Seedance 2.5 support?
480P and 720P at launch, matching the upstream API. Higher tiers are expected to follow as ByteDance expands access.
What should I use in the meantime?
Seedance 2.0 — same model family, same three modes, up to 15 seconds and 4K output, available right now on AutoDirector.
Available right now
Start with Seedance 2.0 while you wait
Same prompting, same modes — your workflow transfers to 2.5 on day one.