Skip to content

Video generation ​

Compatible with OpenAI Videos. Video generation is an asynchronous task: submit, query, download.

ModelToken groupSupported input
grok-imagine-video-1.5grokText-to-video, image-to-video, multi-image reference (up to 14), first and last frames
grok-imagine-video-1.5-litegrokText-to-video, image-to-video

Authentication ​

Authorization: Bearer <token>

Submit a task ​

POSThttps://api.codingrouter.ai/v1/videos

Request parameters ​

ParameterTypeRequiredDescription
modelstringYesSee the table above
promptstringYesVideo description
secondsintegerNoDuration, 1 to 15, default 1
resolutionstringNo480p (default), 720p, 1080p
imagefileNoReference image, form upload; one is image-to-video, several is multi-image reference
first_frame, last_framefileNoFirst and last frames, must be uploaded together, grok-imagine-video-1.5 only
  • When there are images, upload files with multipart/form-data. Image URLs or base64 in JSON are rejected.
  • Multi-image reference and first/last frames do not support 1080p.

Request example ​

bash
curl https://api.codingrouter.ai/v1/videos \
  -H "Authorization: Bearer $CODINGROUTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-imagine-video-1.5",
    "prompt": "Sunset by the sea, the camera slowly pushes in, waves hit the rocks",
    "seconds": 6,
    "resolution": "720p"
  }'
bash
curl https://api.codingrouter.ai/v1/videos \
  -H "Authorization: Bearer $CODINGROUTER_KEY" \
  -F model=grok-imagine-video-1.5-lite \
  -F prompt="Make the person in the picture turn around and smile" \
  -F seconds=6 \
  -F image=@first.png
bash
curl https://api.codingrouter.ai/v1/videos \
  -H "Authorization: Bearer $CODINGROUTER_KEY" \
  -F model=grok-imagine-video-1.5 \
  -F prompt="Transition smoothly from day to night" \
  -F seconds=8 \
  -F first_frame=@day.png \
  -F last_frame=@night.png

Response ​

json
{
  "id": "task_3FnXIsT7uKTvi5r8LFiEDarfYKd1rcGu",
  "object": "video",
  "model": "grok-imagine-video-1.5-lite",
  "status": "queued",
  "progress": 0,
  "created_at": 1794806577
}

Query a task ​

GEThttps://api.codingrouter.ai/v1/videos/{id}
bash
curl https://api.codingrouter.ai/v1/videos/task_3FnXIsT7uKTvi5r8LFiEDarfYKd1rcGu \
  -H "Authorization: Bearer $CODINGROUTER_KEY"
json
{
  "id": "task_3FnXIsT7uKTvi5r8LFiEDarfYKd1rcGu",
  "object": "video",
  "model": "grok-imagine-video-1.5-lite",
  "status": "completed",
  "progress": 100,
  "seconds": 1,
  "resolution": "480p",
  "created_at": 1794806577,
  "completed_at": 1794806621,
  "content_url": "/v1/videos/task_3FnXIsT7uKTvi5r8LFiEDarfYKd1rcGu/content"
}

status goes queued → in_progress → completed, or failed on failure. Query every 5 to 10 seconds. It usually completes in 20 seconds to 1 minute.

Download the video ​

GEThttps://api.codingrouter.ai/v1/videos/{id}/content
bash
curl -L https://api.codingrouter.ai/v1/videos/task_3FnXIsT7uKTvi5r8LFiEDarfYKd1rcGu/content \
  -H "Authorization: Bearer $CODINGROUTER_KEY" \
  -o out.mp4

Returns a video/mp4 file.

Full example (Python) ​

python
import time
import httpx

BASE = "https://api.codingrouter.ai"
headers = {"Authorization": "Bearer sk-your-token"}

with httpx.Client(base_url=BASE, headers=headers, timeout=120) as client:
    task = client.post("/v1/videos", json={
        "model": "grok-imagine-video-1.5-lite",
        "prompt": "Sunset by the sea, the camera slowly pushes in",
        "seconds": 6,
        "resolution": "480p",
    }).json()

    while True:
        time.sleep(8)
        task = client.get(f"/v1/videos/{task['id']}").json()
        if task["status"] in ("completed", "failed"):
            break

    if task["status"] == "completed":
        video = client.get(f"/v1/videos/{task['id']}/content", follow_redirects=True)
        open("out.mp4", "wb").write(video.content)

Billing ​

Billed by the number of seconds generated and the resolution. See the model square for unit prices. Requests that fail parameter validation (400) are not charged.