The Seedance 2.5 API gives you ByteDance's audio-video foundation model for production AI video: native 30-second single-pass clips at up to 1080p, up to 50 multimodal references (30 images, 10 videos, 10 audio clips), first/last-frame control, and video edit and extend for ads, e-commerce, and film previs. It is the same model offered as Dreamina Seedance 2.5.
Generate a 480p draft first, then create a 1080p final video when you are happy with it. Each generation is billed separately.
Explicit types validate task constraints before creation. The prompt must still match the selected type or TaskTypeMismatch may fail asynchronously.
First/last frame tasks require Adaptive. When video or audio references are present, frame roles are treated as reference images.
Upload reference audio
WAV, MP3 (max 15MB, up to 10 audios, duration ≤ 30s)
Video link valid for 72 hours
Transparent pricing with no hidden fees. Pay only for what you use.
| Variant | Tier | GoldCurrent price | Platinum | Diamond | Official | Your savings |
|---|---|---|---|---|---|---|
seedance-2.5 | 480P-input | 0.576Credits(~$0.0576)/secWith reference video: (reference duration + generation duration) × current price | 0.5472Credits(~$0.0547)/sec | 0.5184Credits(~$0.0518)/sec | 0.72Credits(~$0.072)/sec | 20% |
| 480P | 0.9608Credits(~$0.0961)/sec | 0.9128Credits(~$0.0913)/sec | 0.8647Credits(~$0.0865)/sec | 1.201Credits(~$0.1201)/sec | 20% | |
| 720P-input | 1.296Credits(~$0.1296)/secWith reference video: (reference duration + generation duration) × current price | 1.2312Credits(~$0.1231)/sec | 1.1664Credits(~$0.1166)/sec | 1.62Credits(~$0.162)/sec | 20% | |
| 720P | 2.16Credits(~$0.216)/sec | 2.052Credits(~$0.2052)/sec | 1.944Credits(~$0.1944)/sec | 2.7Credits(~$0.27)/sec | 20% | |
| 1080P-input | 3.192Credits(~$0.3192)/secWith reference video: (reference duration + generation duration) × current price | 3.0324Credits(~$0.3032)/sec | 2.8728Credits(~$0.2873)/sec | 3.99Credits(~$0.399)/sec | 20% | |
| 1080P | 5.346Credits(~$0.5346)/sec | 5.0787Credits(~$0.5079)/sec | 4.8114Credits(~$0.4811)/sec | 6.6825Credits(~$0.6683)/sec | 20% |
| Variant | Tier | GoldCurrent price | Platinum | Diamond | Official | Your savings |
|---|---|---|---|---|---|---|
seedance-2.5 | token-input480p / 720p · With reference video | 60Credits/1M tokens~$6/1M tokens | 57Credits/1M tokens~$5.7/1M tokens | 54Credits/1M tokens~$5.4/1M tokens | 75Credits/1M tokens~$7.5/1M tokens | 20% |
token480p / 720p · Without reference video | 100Credits/1M tokens~$10/1M tokens | 95Credits/1M tokens~$9.5/1M tokens | 90Credits/1M tokens~$9/1M tokens | 125Credits/1M tokens~$12.5/1M tokens | 20% | |
token-1080P-input1080p · With reference video | 65.68Credits/1M tokens~$6.568/1M tokens | 62.396Credits/1M tokens~$6.2396/1M tokens | 59.112Credits/1M tokens~$5.9112/1M tokens | 82.1Credits/1M tokens~$8.21/1M tokens | 20% | |
token-1080P1080p · Without reference video | 110Credits/1M tokens~$11/1M tokens | 104.5Credits/1M tokens~$10.45/1M tokens | 99Credits/1M tokens~$9.9/1M tokens | 137.5Credits/1M tokens~$13.75/1M tokens | 20% |
* Actual costs are subject to final output.
* Without a reference video: generation duration × output price. With one: (reference duration + generation duration) × input price.
* Resolution rates are precharge estimates; Auto reserves 30 output seconds. Successful tasks are reconciled using actual token usage.
Generate longer single-pass clips with native audio, dense multimodal references, and a draft-to-final workflow through the Seedance 2.5 API on APIMart.
50K+
Active Users
99.9%
Uptime
2x
Faster
—
Cost Savings
Call ByteDance Seedance 2.5 on APIMart with one POST request. Seedance 2.5 API generation is asynchronous: submit a task, then poll the task endpoint to get the video URL.
Save the key you created above in the APIMART_API_KEY environment variable and send it as Authorization: Bearer <key> on every Seedance 2.5 API request.
POST /v1/videos/generations with model seedance-2.5, your prompt, size, resolution and duration. Add image_urls, video_urls or audio_urls for references. The Seedance 2.5 API responds with a task_id.
Poll GET /v1/tasks/{task_id} every 5–15 seconds until status is completed, then read the video URL from data.result.videos[0].url[0]. A failed Seedance 2.5 API task is fully refunded and returns the reason in data.error.message.
See the Seedance 2.5 API reference for every parameter, task type limit and response field.
curl -X POST https://api.apimart.ai/v1/videos/generations \
-H "Authorization: Bearer $APIMART_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedance-2.5",
"prompt": "A cinematic tracking shot of a red vintage car driving along a coastal road at sunset, waves crashing below, warm golden light",
"size": "16:9",
"resolution": "720p",
"duration": 10
}'import os
import time
import requests
API_BASE = "https://api.apimart.ai"
headers = {"Authorization": f"Bearer {os.environ['APIMART_API_KEY']}"}
# 1. Submit the generation task
payload = {
"model": "seedance-2.5",
"prompt": "A cinematic tracking shot of a red vintage car driving along a coastal road at sunset, waves crashing below, warm golden light",
"size": "16:9",
"resolution": "720p",
"duration": 10
}
resp = requests.post(f"{API_BASE}/v1/videos/generations", headers=headers, json=payload, timeout=60)
resp.raise_for_status()
task_id = resp.json()["data"][0]["task_id"]
# 2. Poll the task until it finishes
for _ in range(200):
task_resp = requests.get(f"{API_BASE}/v1/tasks/{task_id}", headers=headers, timeout=30)
task_resp.raise_for_status()
task = task_resp.json()["data"]
if task["status"] == "completed":
print(task["result"]["videos"][0]["url"][0])
break
if task["status"] == "failed":
raise RuntimeError(task.get("error", {}).get("message", "generation failed"))
time.sleep(10)
else:
raise TimeoutError(f"task {task_id} did not finish in time")curl -X POST https://api.apimart.ai/v1/videos/generations \
-H "Authorization: Bearer $APIMART_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedance-2.5",
"prompt": "The product from the reference image rotates slowly on a white studio turntable, soft key light, the camera pushes in to a close-up of the label",
"image_urls": [
"https://example.com/product.jpg"
],
"size": "16:9",
"resolution": "720p",
"duration": 10
}'TASK_ID="task_xxx" # replace with the task_id returned by the POST request
curl "https://api.apimart.ai/v1/tasks/$TASK_ID" \
-H "Authorization: Bearer $APIMART_API_KEY"| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
model | string | Required | Default: seedance-2.5 | Fixed value for the Seedance 2.5 API: seedance-2.5 (listed upstream as doubao-seedance-2-5). |
prompt | string | Required | Default: — | Video description. Refer to inputs by index, such as @图片1, @视频1 or @音频1 (1-based, in array order). Required for normal and draft generation; do not send it with draft_task_id, or the request returns 400. |
resolution | string | Optional | Default: 720p | 480p, 720p or 1080p. The Seedance 2.5 API returns a 400 error for 2k and 4k. |
size | string | Optional | Default: adaptive | Aspect ratio (also accepted as aspect_ratio): 16:9, 4:3, 1:1, 3:4, 9:16, 21:9 or adaptive. Video edit, video extend and first/last-frame tasks require adaptive. |
duration | integer | Optional | Default: 5 | Seedance 2.5 API clip length: 4–30 seconds, or -1 to let the model choose (precharged as 30 seconds, then settled on the actual output). |
generate_audio | boolean | Optional | Default: true | Generate synced audio (also accepted as audio). Set false for a silent video. |
image_urls | array | Optional | Default: — | Up to 30 Seedance 2.5 API reference images as URLs or asset:// IDs. Use image_with_roles for first and last frames. |
image_with_roles | array | Optional | Default: — | Images with a role: first_frame, last_frame or reference_image. |
video_urls | array | Optional | Default: — | Up to 10 reference videos (mp4 or mov, 2–30 s each, 30 s in total, up to 200 MB each). |
audio_urls | array | Optional | Default: — | Up to 10 reference audio clips (wav or mp3, 2–30 s each, 30 s in total). Audio-only references are supported. |
omni_reference_task_type | string | Optional | Default: auto | auto, reference, edit or extend. Setting edit or extend checks that task's limits when you submit. |
output_format | string | Optional | Default: mp4 | mp4 or mov. mov keeps higher color precision and is recommended for edit and extend. |
draft | boolean | Optional | Default: — | true makes the Seedance 2.5 API generate a 480p draft that stays valid for 7 days. Cannot be combined with draft_task_id. |
draft_task_id | string | Optional | Default: — | Task ID of a completed draft. Renders the 1080p final and inherits the draft's prompt, references, duration and seed. |
return_last_frame | boolean | Optional | Default: false | Also return the last frame image when the task succeeds. |
seed | integer | Optional | Default: — | Random seed. The same seed gives similar, not identical, results. |
watermark | boolean | Optional | Default: false | Add an AI-generated watermark. |
tools | array | Optional | Default: — | Optional tools, such as [{"type": "web_search"}] to reference web information during generation. |
nsfw_check | boolean | Optional | Default: false | Moderate the prompt and input images with omni-moderation-latest before the task is submitted. Video and audio are not checked. |
Compare every Seedance video model on APIMart. One API key calls all of them; each model page shows its live pricing.
| Model | Best for | Resolution | Duration |
|---|---|---|---|
| Seedance 2.5 APIThis page | Clips up to 30 s with up to 50 image, video and audio references, plus video editing and extension | 480P / 720P / 1080P | 4–30 s or auto |
| Seedance 2.0 API | Up to 4K, the highest resolution in the family, with image, video and audio references | 480P / 720P / 1080P / 4K | 4–15 s |
| Seedance 2.0 Fast API | Fast variant of Seedance 2.0 with the same image, video and audio references | 480P / 720P | 4–15 s |
| Seedance 2.0 Mini API | Mini variant of Seedance 2.0 with the same image, video and audio references | 480P / 720P | 4–15 s |
| Seedance 1.5 Pro API | Text-to-video and first/last-frame image-to-video, with optional audio | 480P / 720P / 1080P | 4–12 s |
| Seedance 1.0 Pro Quality API | Text-to-video and first/last-frame image-to-video | 480P / 720P / 1080P | 2–12 s |
| Seedance 1.0 Pro Fast API | Text-to-video and first-frame image-to-video (no last frame) | 480P / 720P / 1080P | 2–12 s |
Guides, comparisons and pricing analysis from the APIMart blog.
ByteDance's newest Seedance model for longer, more controllable AI video production
Production scenarios for 30-second AI video on APIMart
Go from sign-up to your first 30-second clip in three steps
Register on APIMart and create a key on the API Keys page. The same key works for the Seedance 2.5 API and every other model on the platform.
Try prompts, references, duration, and resolution in the playground above before you call the Seedance 2.5 API from code.
Follow the API documentation to submit Seedance 2.5 API tasks from your backend and poll for the finished video.
Example experiences from creator and developer workflows using Seedance 2.5 API
“The visual quality of Seedance 2.5 is incredible! The motion is so smooth and natural, it really elevates my content.”
Sarah Kim
Content Creator
“As a developer, I appreciate the clean API and fast response times. Seedance 2.5 integrates seamlessly into our pipeline.”
Alex Wang
Full-Stack Engineer
Answers about Seedance 2.5 features, parameters, pricing, and access
Seedance 2.5 is ByteDance's audio-video foundation model from the Seed team and the successor to Seedance 2.0. The Seedance 2.5 API uses the model ID seedance-2.5 on APIMart and generates clips of up to 30 seconds at 480p, 720p or 1080p from text, image, video and audio inputs, with sound generated in the same pass. On ByteDance's Volcano Engine the same model is listed as doubao-seedance-2-5; on APIMart, call it with seedance-2.5.
ByteDance announced Seedance 2.5 at the Volcano Engine FORCE conference on June 23, 2026, and its official release date is July 31, 2026. The Seedance 2.5 API is available on APIMart today.
Send POST /v1/videos/generations with model seedance-2.5 and your prompt, then poll GET /v1/tasks/{task_id} until the status is completed and read the video URL from data.result.videos[0].url[0]. The Seedance 2.5 API example on this page has cURL and Python code, and the full Seedance 2.5 API reference is on docs.apimart.ai.
Seedance 2.5 doubles the maximum length from 15 to 30 seconds and raises reference limits from 9 images, 3 videos and 3 audio clips to 30 images, 10 videos and 10 audio clips. It also adds audio-only references, mov output, a 480p draft mode and a watermark option. The Seedance 2.5 API on APIMart outputs up to 1080p; for 4K output, use the Seedance 2.0 API. The Seedance 2.5 API and the Seedance 2.0 API work with the same APIMart API key.
In the Seedance 2.5 API, model (seedance-2.5) and prompt are required; prompt must not be sent with draft_task_id. Optional parameters are resolution (480p, 720p or 1080p), size (16:9, 4:3, 1:1, 3:4, 9:16, 21:9 or adaptive), duration (4–30 or -1), generate_audio, image_urls, image_with_roles, video_urls, audio_urls, omni_reference_task_type, output_format (mp4 or mov), draft, draft_task_id, return_last_frame, seed, watermark, tools and nsfw_check. The Seedance 2.5 API parameter table above lists defaults and limits.
The Seedance 2.5 API generates 4 to 30 seconds per request, or set duration to -1 to let the model choose; if you omit duration, the clip is 5 seconds. Seedance 2.5 API resolution is 480p, 720p (default) or 1080p. 2k and 4k are not supported.
Up to 50 per Seedance 2.5 API request: 30 images, 10 videos and 10 audio clips. Reference videos and reference audio can each total up to 30 seconds, and audio-only references are supported. Real human faces must be uploaded to the asset library first and referenced as asset:// IDs.
Yes. The Seedance 2.5 API supports text to video, image to video with reference images, first-frame and first-and-last-frame video, video editing and video extension. Edit, extend and frame tasks require size adaptive, and edits also require duration -1 with a 4–30 second source clip. In the Seedance 2.5 API, set omni_reference_task_type to edit or extend to catch these limits when you submit.
Yes. The Seedance 2.5 API generates video and audio together, and generate_audio is true by default. Set it to false for a silent clip. You can also pass audio_urls as music or voice references.
For the Seedance 2.5 API, describe one continuous shot with the subject, action, camera movement and lighting, and refer to inputs by index, such as @图片1 for the first image or @视频1 for the first video. For edits, say what to add, remove or replace; for extensions, say to extend forward or backward. Keep edit words out of normal reference prompts, or the task may be treated as an edit.
The Seedance 2.5 API is settled by actual token usage. The per-second rates on this page, by resolution and with or without reference video, are precharge estimates; successful tasks are reconciled against the tokens actually used, and failed tasks are fully refunded. The Seedance 2.5 API pricing table shows Gold member prices (20% off); Platinum and Diamond members save up to 28%.
No. There is no free quota for the Seedance 2.5 API on APIMart. Billing is pay-as-you-go and you only pay for successful Seedance 2.5 API requests. Members save 20%–28%: Gold members get 20% off, and Platinum and Diamond members up to 28%. See this page's live pricing table for current prices.
Sign up for APIMart, open the API Keys page in your dashboard and create a new key. The same key calls the Seedance 2.5 API and every other model on the platform, with pay-as-you-go billing and no subscription.
Dreamina Seedance 2.5 is how ByteDance's Dreamina app names the Seedance 2.5 model. In Dreamina you use it inside the app; on APIMart you call it through the API with the model ID seedance-2.5. You may also see it written as Seedance2.5 or ByteDance Seedance 2.5.
Explore more models in the same category.

Flux 3 Video
FLUX 3 Video by Black Forest Labs supports text-to-video, image-to-video with ordered keyframes and video continuation. It generates 5–20 second HD or FHD clips with synchronized audio in seven aspect ratios.

MiniMax H3
MiniMax H3 is MiniMax's video model for text-to-video, image-to-video with first and last frames, and multimodal reference-to-video using images, videos and audio. It outputs 4–15 second clips at 2K or 768P with an audio track.

Gemini Omni Flash Preview
gemini-omni-flash-preview is a multimodal video generation and editing model launched by Google.

Seedance 2.0 Mini
Seedance 2.0 Mini is the lightweight version of ByteDance's Seedance 2.0 for text-to-video and image-to-video, with 480p or 720p output.