Generate long-form cinematic AI video with ByteDance's Seedance 2.5 through one unified API. Single-pass clips up to 30 seconds with native synchronized audio, up to 50 multimodal references, and 4K output — behind one REST endpoint with async jobs and webhooks, and no separate ByteDance account to manage.
Integrate Seedance 2.5 with a single API call — one key, one unified endpoint, and shared billing across every model on Apiframe.
model: "seedance-2.5" Coming soon Seedance 2.5 generates a single native clip up to 30 seconds in one pass, with scene changes and tempo shifts built into the same generation. Because the whole shot holds one continuous context, motion, lighting, and identity stay coherent instead of drifting at each splice point the way stitched clips do.
A single generation can take up to 50 multimodal references — reference images, short video clips, and audio — a big jump from Seedance 2. Tag each input with a role so the model knows which one should pin down character identity, product look, brand palette, camera style, or voice.
It renders natively at up to 4K (3840x2160) with 10-bit color for smoother gradients and more room in post. Audio is co-generated in the same pass rather than dubbed on afterward, so music, sound effects, dialogue, and lip-sync land synchronized to the picture.
Describe multi-shot sequences and camera direction — pans, dollies, orbits, focus changes — directly in the prompt, and Seedance 2.5 keeps the subject consistent across cuts. One call returns an edited-feeling sequence rather than a single take, with roughly 20% better prompt adherence than Seedance 2.
Edit a specific region or element while the rest of the frame stays visually stable, or edit an existing clip as a reference video: keep its character, environment, camera movement, composition, action rhythm, and duration, and change only what you ask for — no regenerating the whole take.
It holds a character's identity steady across movement, angle changes, and scene transitions, which matters for films, brand spots, and serialized content. Native audio and on-screen text render across many languages, so one generation can localize without a separate pass.
A few outputs generated through the Seedance 2.5 API on Apiframe.
Drone shot flying over a tropical coastline at sunrise, smooth cinematic motion
A vintage car driving through a rainy neon city at night, reflections on the asphalt
Time-lapse of clouds rolling over a mountain range in warm golden light
Send a single POST /v2/videos/generate request with your API key to
generate with Seedance 2.5. The call returns a jobId you can poll or
receive via webhook.
curl -X POST https://api.apiframe.ai/v2/videos/generate \
-H "X-API-Key: afk_your_api_key_here" \
-H "Content-Type: application/json" \
-d '{
"prompt": "a cinematic sunrise over a futuristic cityscape, smooth camera push-in",
"model": "seedance-2.5",
"seedanceParams": {
"resolution": "720p",
"start_image": "https://example.com/input.jpg",
"generate_audio": false,
"end_image": "https://example.com/input.jpg"
}
}'import requests
response = requests.post(
"https://api.apiframe.ai/v2/videos/generate",
headers={
"X-API-Key": "afk_your_api_key_here",
"Content-Type": "application/json",
},
json={
"prompt": "a cinematic sunrise over a futuristic cityscape, smooth camera push-in",
"model": "seedance-2.5",
"seedanceParams": {
"resolution": "720p",
"start_image": "https://example.com/input.jpg",
"generate_audio": False,
"end_image": "https://example.com/input.jpg"
}
},
)
print(response.json()) # { "jobId": "...", "status": "QUEUED" }const response = await fetch("https://api.apiframe.ai/v2/videos/generate", {
method: "POST",
headers: {
"X-API-Key": "afk_your_api_key_here",
"Content-Type": "application/json",
},
body: JSON.stringify({
"prompt": "a cinematic sunrise over a futuristic cityscape, smooth camera push-in",
"model": "seedance-2.5",
"seedanceParams": {
"resolution": "720p",
"start_image": "https://example.com/input.jpg",
"generate_audio": false,
"end_image": "https://example.com/input.jpg"
}
}),
});
const { jobId } = await response.json();
console.log(jobId);Generation is asynchronous. A successful submission returns 202 Accepted with a jobId. Poll GET /v2/jobs/{id} (or supply a webhook_url) until the status is COMPLETED; the result field then holds the output URL(s).
{
"jobId": "b2c3d4e5-f6a7-8901-bcde-f23456789012",
"status": "QUEUED"
}curl https://api.apiframe.ai/v2/jobs/JOB_ID \
-H "X-API-Key: afk_your_api_key_here"import requests, time
while True:
job = requests.get(
"https://api.apiframe.ai/v2/jobs/JOB_ID",
headers={"X-API-Key": "afk_your_api_key_here"},
).json()
if job["status"] in ("COMPLETED", "FAILED"):
break
time.sleep(2)
print(job["result"])let job;
do {
await new Promise((r) => setTimeout(r, 2000));
job = await fetch("https://api.apiframe.ai/v2/jobs/JOB_ID", {
headers: { "X-API-Key": "afk_your_api_key_here" },
}).then((r) => r.json());
} while (job.status !== "COMPLETED" && job.status !== "FAILED");
console.log(job.result);Request parameters accepted by the Seedance 2.5 endpoint. Model-specific options are nested under the params object shown below.
| Parameter | Type | Required | Default | Allowed / range | Description |
|---|---|---|---|---|---|
| prompt | string | required | — | — | Text description of what to generate. |
| model | string | required | "seedance-2.5" | "seedance-2.5" | The model identifier for this endpoint. |
| seedanceParams.resolution | string | optional | "720p" | "480p", "720p", "1080p", "4k" | Resolution |
| seedanceParams.start_image | string (URL) | optional | — | — | Use a still as the first frame of the clip. |
| seedanceParams.generate_audio | boolean | optional | false | — | Generate audio |
| seedanceParams.end_image | string (URL) | optional | — | — | Optional last frame to control the ending. |
| seedanceParams.reference_image_urls | string[] (URLs) | optional | — | — | Reference images |
| seedanceParams.reference_video_urls | string[] (URLs) | optional | — | — | Reference videos |
| seedanceParams.reference_audio_urls | string[] (URLs) | optional | — | — | Reference audio |
| seedanceParams.return_last_frame | boolean | optional | false | — | Also return the final frame as a still image. |
| seedanceParams.web_search | boolean | optional | false | — | Web search |
| seedanceParams.camera_fixed | boolean | optional | false | — | Disable camera movement. |
| seedanceParams.seed | number | optional | — | step 1 | Reuse a number to reproduce the same result. |
Common questions about the Seedance 2.5 API.
Seedance 2.5 is ByteDance's next-generation multimodal AI video model and a generational leap from Seedance 2.0, announced in June 2026 with general availability in early July 2026. Its headline features are single-pass 30-second clips, up to 50 multimodal references, native 4K output, region-level local editing, director-grade camera control, and native synchronized audio.
Text, image, audio, and video — up to 50 multimodal references combined in a single generation. Tag each input with a role to guide character, style, motion, camera, or sound.
Yes. Audio is co-generated in the same latent space as the picture, producing synchronized native audio — music, sound effects, dialogue, and lip-sync — instead of a silent clip you score afterward.
Up to 30 seconds in a single pass, at up to 4K (3840x2160) with 10-bit color, across multiple aspect ratios. Pricing is per second of output.
Yes. It supports region-level local editing — change one element while the rest of the frame stays consistent — and video-reference editing that preserves the source clip's character, environment, camera movement, composition, action rhythm, and duration while swapping or adding what you specify.
Per second of generated video, so a longer clip simply costs proportionally more — you only pay for successful generations.
Still have questions?
Get your API key and integrate Seedance 2.5 in minutes — Pay-as-you-go.
Questions? Join our Discord or contact sales.