Up to 30 seconds in one take
Twice the length of Seedance 2.0. A whole tutorial, a monologue or a short scene comes out as one continuous performance instead of stitched clips.
Seedance 2.5 · by ByteDance
Seedance 2.5 AI video generator
Seedance 2.5 is ByteDance's newest video model. It runs up to 30 seconds in one generation, with dialogue, lip sync and sound made in the same pass, and lets you draft a scene cheaply before rendering it in full. Write a scene below and it opens in Video Studio on Seedance 2.5.
Example video
Opens in Video Studio with the model you picked. You see the price in credits before anything runs.
What Seedance 2.5 is known for
Dialogue and sound in the same pass
Put lines in double quotes and they are spoken and lip-synced, with room tone, effects and music generated alongside. Sound costs nothing extra.
Draft first, then finish
Draft mode renders a cheap low-resolution preview. Keep the one you like and enhance it to full quality, so you only pay full price for takes you keep.
More references at once
Image to video and Reference to video take your own frames, characters and products. ByteDance says 2.5 follows prompts about 20% better than 2.0.
Where it's weaker
- It costs more per second than Seedance 2.0, and there is no Fast tier.
- Fast, wide action can smear or morph; it is at its best with people talking and moving at human speed.
- Faces soften a little in tight push-ins while someone speaks.
- Real people's faces are often refused as references; generated or illustrated characters work.
- Render times vary a lot, from about a minute to several.
- Prompts written for 2.0 sometimes need rewriting; 2.5 resists very fast cuts.
- Maker
- ByteDance Seed · unveiled June 2026, wide release July–August 2026
- Clip length on Affogato
- Any length from 4 to 30 s
- Resolution on Affogato
- 480p · 720p · 1080p; Draft previews at 480p
- Aspect ratios
- 9:16 · 3:4 · 1:1 · 4:3 · 16:9 · 21:9
- Sound
- Dialogue, lip sync, effects and music, on by default at no extra cost
- Inputs
- Text · start frame · reference images of up to four characters
Made with Seedance 2.5 on Affogato
Every clip below was generated with the prompt shown, unedited. Load any prompt into the generator and make your own version.
20-second take
Grandma's seven pleats
Prompt
One continuous 20-second vertical take, phone on a tripod in a small sunny kitchen. A grandmother in her seventies with silver hair in a bun and a flour-dusted blue apron teaches her grandson's followers how to fold dumplings. She looks into the camera and says, "First, a little water on the edge, like this." She wets the wrapper with her fingertip, places a spoon of filling, then pinches seven quick pleats while saying, "Seven pleats. Not six. Seven." She holds the finished dumpling up to the lens, laughs and says, "Your turn." Natural daylight, steam from a pot behind her. Sound: her warm voice, a kettle simmering, a radio playing quietly in another room.
Creator to camera
Get ready with me, interview day
Prompt
Vertical get-ready-with-me video filmed on a phone propped on a bathroom counter. A young woman in her early twenties with a messy bun and a white T-shirt applies blush with her fingers while talking to the camera: "So I have a job interview in an hour and I am not panicking." She pauses, looks at the lens and adds, "I am panicking a little." She laughs and blends the blush, then picks up a lip tint. Soft vanity light, real skin texture, slight phone wobble. Sound: her voice, the clink of makeup on the counter.
Two speakers
Podcast hot take
Prompt
A video podcast in a cozy studio with warm lamps and acoustic panels. Two hosts sit at a wooden desk with microphones on arms. The woman on the left, late twenties with braids and a green sweater, says, "Okay, hot take: breakfast for dinner is the best dinner." The man on the right, thirties, beard and glasses, leans back, laughs and says, "That's not a hot take, that's a fact." She points at him, laughing. Cut to a close-up of him shaking his head, smiling. Sound: their natural voices close to the mics, a soft chair creak.
Acting
A proposal in the park
Prompt
Vertical handheld video in a park at golden hour. A young man in a denim jacket kneels in front of his girlfriend on a path lined with autumn trees and opens a small ring box. She freezes, covers her mouth with both hands, her eyes fill with tears, and she nods fast, laughing, "Yes! Yes, obviously yes!" She pulls him up into a hug while strangers on a nearby bench clap. Warm low sunlight, falling leaves. Sound: her voice breaking, applause, birds.
Fashion, two shots
Emerald silk on the runway, then backstage
Prompt
Vertical fashion clip in two shots. Shot one: a tall model with a shaved head in a flowing oversized emerald silk suit walks the runway toward the camera under hard white light, fabric rippling with each step, the audience a dark blur. Shot two: backstage, the same model drops his runway face, grins at a friend off-camera and flicks the lapel with a laugh. Sound: a deep electronic runway beat, camera shutters, then backstage chatter.
UGC product review
The noise cancelling review
Prompt
Vertical UGC product review shot on a phone: a man in his thirties with a short beard sits on his couch, opens a matte black box and lifts out a pair of over-ear headphones. He puts them on, his eyebrows shoot up, and he says to the camera, "Okay. The noise cancelling is actually unreal." He takes them off and taps the ear cup. Living-room daylight, natural skin, slight handheld movement. Sound: cardboard sliding, his voice, room tone dropping out for a moment when the headphones go on.
Group motion
Underpass dance crew
Prompt
Vertical clip of a street-dance crew of four in a concrete underpass at night, lit by orange streetlights. They hit a sharp synchronized routine: a chest pop, a quick spin, a freeze on the beat, then the woman in the middle breaks out into a solo footwork sequence while the others cheer. Loose streetwear, sneakers squeaking. The camera dollies slowly back. Sound: a punchy hip-hop beat, sneakers, the crew shouting encouragement.
Which version to pick
Start here
Seedance 2.5
Long takes and dialogue
Anything that has to run past 15 seconds, or a scene where people talk. Use Draft to try the take first.
5 s clip 429 credits · 4–30 s
Open with Seedance 2.5Option 2
Seedance 2.0
Clips up to 15 seconds
The same family at a lower price per second. For short clips it is usually the better buy.
5 s clip 275 credits · 4–15 s
Open with Seedance 2.0Option 3
Seedance 2.0 Fast
Quick drafts
Tuned for speed at a lower price. Find the shot here before spending on 2.5.
5 s clip 220 credits · 4–15 s
Open with Seedance 2.0 Fast| Seedance 2.5 | Seedance 2.0 | |
|---|---|---|
| Released | June 2026 (wide release July) | February 2026 |
| Clip length | 4–30 s | 4–15 s |
| Draft, then enhance | Yes | No |
| Fast tier | No | Yes |
| References | More at once (ByteDance says up to 50) | Up to 9 images, or clips |
| Prompt following | About 20% better, per ByteDance | Baseline |
| Price for the same length | Higher | Lower |
For a clip of 15 seconds or less, Seedance 2.0 costs less and looks close. Choose 2.5 when the take has to run longer, when people talk for most of it, or when you want to draft cheaply first.
Seedance 1.5 Pro and 1.0 Pro are still in the studio's model menu: much cheaper, shorter clips, and 1.0 has no sound. Credits are read from the live catalog; the Standard plan ($24/month) includes 2,000 credits, about 4 five-second Seedance 2.5 clips.
Prompting Seedance 2.5 well
- Write it as one scene: who is in frame, what they do and say in order, then the place, light and camera.
- Put every spoken line in double quotes and keep each line short. Say how it is said ("laughs and says", "whispers").
- Name the accent when it matters: "American" works better than "American English".
- For long takes, give the performer three or four beats of action; 2.5 fills the time between them.
- Describe two or three shots at most ("Shot one:", "Shot two:"); it resists rapid cutting.
- Name the sounds you want and write "no music" if you don't want a score.
- Draft the take, check timing and performance, then enhance the keeper to full quality.
From prompt to clip
Describe the scene or add a photo
Write the take in the box at the top of this page, or start from an example. Add a photo to use it as the first frame.
Sign in and land in Video Studio
Your prompt, photo, format and the Seedance version you picked carry over. Nothing runs until you press Generate.
Pick the length, or draft first
Choose any length from 4 to 30 seconds. Turn on Draft for a cheap preview; the credit price shows before anything is spent.
Enhance the keeper
When a draft works, press Enhance to render it at full quality. Takes you don't keep never cost full price.
Questions about Seedance 2.5
What is Seedance 2.5 best at?
Long, continuous takes of people talking and moving: tutorials, monologues, creator videos and short scenes with dialogue, up to 30 seconds with the sound made in the same pass. It is weaker on very fast action.
What is the difference between Seedance 2.5 and 2.0?
2.5 runs up to 30 seconds instead of 15, adds a Draft mode, takes more references at once and follows prompts more closely. It costs more per second and has no Fast tier, so 2.0 is the cheaper choice for short clips.
How does Draft mode work?
A draft is a quick low-resolution render that costs much less. If you like it, Enhance renders the same take at full quality; if not, you change the prompt and draft again.
Can Seedance 2.5 make 4K video?
Not on Affogato. ByteDance's 4K build is a limited trial for selected film studios; the public model renders up to 1080p, here as everywhere else.
Can I use a photo of myself?
Often not. ByteDance blocks most references that show real faces unless the person has gone through its verification. Generated or illustrated characters, products and places work.
Is Seedance 2.5 free on Affogato?
No. Every run costs credits, priced by the second and shown before you generate. Credits come with a monthly plan and unused credits roll over up to the plan's cap.
Can I use Seedance 2.5 videos commercially?
Yes. Paid Affogato plans include a commercial license and no watermark, within our terms of service and ByteDance's content rules.
Keep exploring
Facts checked November 15, 2026. Sources: ByteDance Seed: Seedance 2.5 · BytePlus: Seedance 2.5 launch · The Next Web: Seedance 2.5 unveiled · TechNode: Seedance 2.5 rollout · fal: Seedance 2.5 vs 2.0 · fal: Seedance 2.5 API · Segmind: Seedance 2.5 review · MindStudio: Seedance 2.5 review