# Seedance 2.5 Upscaled (`bytedance/seedance-2.5-upscaled`): API

> bytedance/seedance-2.5-upscaled is a video generation model from ByteDance you can call on the OceanToken API.

Endpoint: `POST https://api.oceantoken.ai/v1/videos`. Auth: `Authorization: Bearer $OCEANTOKEN_API_KEY`. OpenAI-compatible; the OpenAI SDKs work with base URL https://api.oceantoken.ai/v1.

The call creates a job: poll `GET /v1/videos/{video_id}`, then download `GET /v1/videos/{video_id}/content`.

## Request parameters

- `model` (string; required): This model's ID.
- `prompt` (string; required): What happens in the clip.
- `duration` (integer; 4 to 15; default 5): Length in whole seconds. `seconds` is accepted too.
- `resolution` (string; 720p, 1080p): Output resolution; lower is faster and cheaper.
- `aspect_ratio` (string; 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, adaptive): Output aspect ratio. `adaptive` follows the first frame (the default when one is sent and no ratio is). If the request is served where `adaptive` is not available it is ignored and named in the `x-oceantoken-dropped-params` response header.
- `generate_audio` (boolean): `true` for a clip with sound, `false` for a silent one.
- `frame_images` (array; up to 2 items): The first frame and last frame: `{"type": "image_url", "frame_type": "first_frame", "image_url": {"url": ...}}`.
- `omni_reference_task_type` (string; auto, reference, edit, extend): What a request with reference media asks for: `reference` builds a new clip from them, `edit` changes the reference video, `extend` continues it, `auto` lets the model decide from the prompt. Say the same in the prompt (for example "continue this video"): the model also goes by the prompt. If the request is served where this is not available it is ignored and named in the `x-oceantoken-dropped-params` response header. Left out, the model decides from the prompt.
- `input_references` (array): Reference media the clip is built from: up to 9 images, up to 10 videos, up to 10 audios. Each `{"type": "image_url" | "video_url" | "audio_url", ...: {"url": ...}}`.

## Example: Text to video

```bash
curl https://api.oceantoken.ai/v1/videos \
  -H "Authorization: Bearer $OCEANTOKEN_API_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "bytedance/seedance-2.5-upscaled",
  "prompt": "A paper boat drifting down a rain-soaked street",
  "duration": 5,
  "resolution": "720p"
}'
# then poll GET https://api.oceantoken.ai/v1/videos/{id} until "status" is "completed"
```

```python
import os, time, uuid
import requests

BASE = "https://api.oceantoken.ai/v1"
headers = {"Authorization": f"Bearer {os.environ['OCEANTOKEN_API_KEY']}"}
job = requests.post(
    f"{BASE}/videos",
    headers={**headers, "Idempotency-Key": str(uuid.uuid4())},
    json={
        "model": "bytedance/seedance-2.5-upscaled",
        "prompt": "A paper boat drifting down a rain-soaked street",
        "duration": 5,
        "resolution": "720p",
    },
).json()
while job["status"] in ("queued", "in_progress"):
    time.sleep(15)
    job = requests.get(f"{BASE}/videos/{job['id']}", headers=headers).json()
print(job["status"], job.get("content_url"))  # the MP4, valid for 24 hours
```

## Response

- `id` (string): The job's ID; poll it.
- `status` (string; queued, in_progress, completed, failed): Where the job is.
- `seconds` (string): Length of the clip.
- `size` (string): Width x height of the clip.
- `content_url` (string): Once completed: the MP4, valid for 24 hours.
- `error` (object): On a failed job: `code` and `message`.

```json
{
  "id": "video_...",
  "object": "video",
  "status": "completed",
  "model": "bytedance/seedance-2.5-upscaled",
  "seconds": "5",
  "size": "1280x720",
  "created_at": 1792000000,
  "completed_at": 1792000095,
  "content_url": "https://...",
  "error": null
}
```

## More

- Model page: https://docs.oceantoken.ai/models/bytedance/seedance-2.5-upscaled
- This spec as JSON (with a JSON Schema of the request): https://docs.oceantoken.ai/models/bytedance/seedance-2.5-upscaled/spec.json
- Guide: https://docs.oceantoken.ai/guides/video
- All models: https://docs.oceantoken.ai/llms.txt