ByteDance · New on Visiojoy, October 2026 · Successor to Seedance 2.0
Make videos up to 30 seconds long, with sound, from a text prompt, a starting image or your own references. Write the scene below and generate it in your browser — nothing to install.
Seedance 2.5 is ByteDance's newest Seedance video model and the next step after Seedance 2.0. It keeps what made 2.0 popular — stable motion, sound generated with the picture, and image, video and audio references — and doubles the clip length to 30 seconds.
That extra length changes what one generation can hold: a scene with a setup and a payoff, a full 30-second ad, or a conversation between two characters. On Visiojoy you can run it from a text prompt, from a start (and end) frame, or from a mix of references, in 6 aspect ratios from vertical 9:16 to cinematic 21:9, plus Auto. Visiojoy is an independent platform, not affiliated with ByteDance.
Four things that change how you plan a clip.
Seedance 2.0 stops at 15 seconds. Seedance 2.5 doubles that to 30, so a full product spot, a short scene with dialogue or a slow establishing shot fits in one clip instead of three stitched ones. Duration is set in one-second steps from 4 to 30.
Audio is generated together with the video and is on by default: ambient sound, effects, music and spoken lines written in quotes. Turn it off when you plan to score the clip yourself.
Feed it up to 9 images, 3 videos and 3 audio clips in one request — a character sheet, a camera move to copy, a voice or a track to sync to. Reference video and audio can each run up to 30 seconds in total, twice the Seedance 2.0 allowance.
Upload a start image to keep a product, face or layout exact from the first frame. In the full AI Video workspace you can also add an end frame so the clip lands where you need it — useful for transitions and loops.
Three ways to start a clip. The generator above switches mode based on what you upload; end frames and the full settings panel are in the AI Video workspace.
You upload
Nothing — just a prompt
Describe the subject, action, camera and sound. Best for exploring ideas and scenes you don't have assets for yet.
You upload
1 start image, optional end image
The clip opens on your image. Add the end frame in the AI Video workspace to control where it finishes. Best for product shots, character consistency and transitions.
You upload
Up to 9 images, 3 videos, 3 audio clips
Mix visual, motion and sound references and describe how to use them in the prompt. Can't be combined with a first frame.
Seedance 2.5 prompts written for longer clips and built-in audio. Click one to load it into the generator, then match the settings shown on the card.
Longer clips leave room for a beginning, a turn and an ending. Write the scene as a sequence of beats and Seedance 2.5 plays them in order.
An old lighthouse keeper climbs a spiral staircase with a lantern, reaches the top, and lights the lamp as a storm rolls in over the sea. Beat 1: slow climb, close on his hands on the rail. Beat 2: he opens the door, wind rushes in. Beat 3: wide shot, the beam sweeps across black waves. Sound: creaking stairs, howling wind, thunder in the distance.
A six-year-old girl in a yellow raincoat stands at the school gate, grips her backpack straps, takes a deep breath, then runs inside to join a group of kids. Handheld camera at child height, soft morning light. Sound: playground chatter, a bell ringing, her sneakers on wet pavement.
A red vintage convertible drives along an empty desert highway at golden hour. Start with a drone shot from high above, descend to follow the car from behind, then cut to the driver's sunglasses reflecting the sunset, end on the car shrinking toward the horizon. Sound: engine hum, wind, a classic rock song on the car radio.
Use one clip for a full spot: hook, product moment and closing shot. Upload a product photo as the first frame to keep packaging and colors exact.
A white running sneaker drops onto a wet rooftop at night and splashes water in slow motion, camera orbits around it as city lights blur behind, then cuts to a runner sprinting past in the same shoes. Bold, punchy pacing. Sound: a heavy bass hit on impact, splashing water, fast footsteps.
Macro shot of a glass serum bottle on a marble tray, a dropper releases one golden drop onto fingertips, slow push-in on skin texture, soft daylight from the left. Calm, premium mood. Sound: a gentle drip, soft fabric rustle, quiet ambient piano.
A courier rides through a busy evening street, hands over a steaming pizza box at an apartment door, the box opens to reveal melting cheese pulling apart in close-up. Energetic cuts. Sound: scooter engine, doorbell, a friendly 'Enjoy!' from the courier, cheese sizzling.
Audio is generated with the picture. Put spoken lines in quotes and name the sounds you want; add a reference audio clip when the voice or music must match.
Two friends sit across from each other in a small diner booth at night. The woman leans in and says: "You really quit your job?" The man smiles and says: "Tomorrow I start building boats." Slow push-in, warm neon from the window. Sound: diner chatter, coffee cups clinking.
A street musician plays acoustic guitar and sings softly under a bridge at dusk, a small crowd gathers, a child drops a coin into the guitar case. Camera slowly circles the musician. Sound: fingerpicked guitar, a gentle folk melody, light applause at the end.
A friendly science host stands in a bright studio next to a large globe and explains to camera: "Every day, the Earth spins about forty thousand kilometers. Here's why you don't feel it." She spins the globe, the camera pushes in on its surface. Clear voice, light background music.
All three are in the same model menu. Here's when to switch.
| Seedance 2.5 | Seedance 2.0 | Seedance 2.0 Fast | |
|---|---|---|---|
| Clip length | 4–30s | 4–15s | 4–15s |
| Resolution | 480p – 1080p | 480p – 4K | 480p – 720p |
| Reference video / audio, total | 30s / 30s | 15s / 15s | 15s / 15s |
| 720p cost | 63 credits / s | 41 credits / s | 25 credits / s |
| Pick it for | Longer scenes, dialogue, finished spots | High-resolution short clips, up to 4K | Cheap drafts and quick iteration |
Comparing other providers? See Wan 3.0 for cheaper long clips, MiniMax H3 for 2K, or browse all AI video models. Building an agent workflow? Seedance 2.5 is also available through the Visiojoy MCP server.
You pay per second of video. The exact cost shows on the Generate button before you start.
| Resolution | Per second | 5s clip | 15s clip | 30s clip |
|---|---|---|---|---|
| 480p | 28 credits | 140 | 420 | 840 |
| 720p | 63 credits | 315 | 945 | 1,890 |
| 1080p | 158 credits | 790 | 2,370 | 4,740 |
Credit packs start at $9.90 for 640 credits — about 10 seconds of 720p Seedance 2.5 video. With reference videos, billing switches to a lower per-second rate (480p 17 · 720p 38 · 1080p 95) applied to the reference length plus the output length. Image and audio references don't change the price. See plans and credit packs.
1
Subject, action, camera move, lighting — then the sound. Put any spoken lines in quotes. For longer clips, list the beats in order.
2
Use + to set a first frame, or switch to Reference to Video to add up to 15 images, videos and audio clips. Images are optional — a prompt alone works.
3
Pick 4–30 seconds, an aspect ratio and 480p, 720p, 1080p, then generate. The credit cost shows before you start, and download the finished video from the AI Video workspace.
Seedance 2.5 is a video generation model from ByteDance and the successor to Seedance 2.0, available on Visiojoy since October 2026. It generates clips of 4 to 30 seconds at 480p, 720p, 1080p with audio generated alongside the video. On Visiojoy you can use it for text-to-video, first and last frame video, and reference-to-video with image, video and audio inputs.
The biggest change is length: Seedance 2.5 makes clips up to 30 seconds, compared with 15 seconds on Seedance 2.0, and accepts up to 30 seconds each of reference video and audio instead of 15. Seedance 2.0 still offers 4K output, while Seedance 2.5 tops out at 1080p.
Any length from 4 to 30 seconds, set in one-second steps before you generate. The default is 5 seconds.
No. Seedance 2.5 outputs at 480p, 720p, 1080p. If you need 4K, use Seedance 2.0, which goes up to 4K in clips of up to 15 seconds.
Yes. Audio is generated together with the video and is switched on by default. Describe the sounds, music or dialogue you want in the prompt — put spoken lines in quotes. You can switch audio off in the settings if you only need the picture.
In Reference to Video mode you can add up to 9 images (JPG, PNG or WebP, up to 30 MB each), 3 videos (MP4 or MOV, up to 100 MB each, 30 seconds in total) and 3 audio clips (MP3, WAV, AAC, M4A or OGG, up to 15 MB each, 30 seconds in total). Adding references requires signing in.
Yes. Upload an image as the first frame and the video starts from it; in the AI Video workspace you can also add a last frame for the clip to end on. First and last frames can't be combined with multi-modal references — if you need an exact opening shot plus other references, add that image as a reference and describe its role in the prompt.
It is billed per second of output: 28 credits per second at 480p, 63 at 720p and 158 at 1080p. A 5-second 720p clip costs 315 credits; a 30-second 1080p clip costs 4740. When you add reference videos, the reference length is billed too, at a lower rate: 17 / 38 / 95 credits per second of reference plus output.
Not entirely. You can write prompts and set up a generation without an account, and new accounts get 12 free credits, but the shortest Seedance 2.5 clip (4 seconds at 480p) costs 112 credits. To generate, you need a credit pack (from $9.90 for 640 credits) or a subscription (from $15/month). For cheaper drafts, Seedance 2.0 Fast costs 12 credits per second at 480p.
All three generate video with sound and accept image, video and audio references. Seedance 2.5 is the polished choice for scenes up to 30 seconds at 63 credits per second for 720p. Wan 3.0 also reaches 30 seconds and costs less (16 credits per second at 720p). MiniMax H3 is the pick for 2K resolution and stereo sound in clips up to 15 seconds.
Yes. On Visiojoy you can use Seedance 2.5 from your browser anywhere the site is available. You sign in with your Google account — no Chinese phone number or Douyin account needed.
No. ByteDance hasn't released model weights for Seedance 2.5 or any earlier Seedance model, so it can't be downloaded or run locally. It's available through ByteDance's own apps and APIs, and through platforms like Visiojoy.
ByteDance offers Seedance models to developers through its BytePlus and Volcano Engine APIs. Without setting up a separate account, you can also generate Seedance 2.5 videos from Claude, Cursor and other MCP clients through the Visiojoy MCP server, using the same credits as on this website.
No. Visiojoy is an independent AI video platform and is not affiliated with ByteDance. It offers Seedance 2.5 alongside Seedance 2.0, MiniMax H3, Wan 3.0, Grok Imagine and other video models in one account.
Yes, once you've made a purchase. Videos created after your first purchase (any subscription or credit pack) can be used commercially; videos created before your first purchase are licensed for personal, non-commercial use only. Also check the destination platform's policy before publishing.
Write a scene, add a start frame or references, and generate a Seedance 2.5 video with sound — all in the browser.
