Skip to main content
OpenAI’s text-to-video flagship. Async by nature: submit returns a task_id, then poll tasks/query until the result lands. Typical latency 60–120 seconds for a 5-second clip. Slug: sora-2 · Capability: 1080p text-to-video, up to 10s · Pricing: per-second (live rate)

​
Submit

curl https://llm.bytespike.ai/v1/tasks/submit \
  -H "x-api-key: $BYTESPIKE_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "sora-2",
    "prompt": "A drone shot tracking a yellow cab through Times Square at golden hour.",
    "duration_seconds": 5,
    "resolution": "1080p"
  }'

​
Body

FieldTypeRequiredNotes
modelstringyessora-2
promptstringyesText prompt.
duration_secondsintegeryes1–10
resolutionstringno720p (default), 1080p
aspect_ratiostringno16:9 (default), 9:16, 1:1
seedintegernoReproducibility.
image_initstringnoURL of init image for image-to-video.

​
Response

{
  "task_id": "task_01AbCdEf",
  "status": "queued",
  "estimated_credits": 0.42
}
The estimated credits land before any GPU work runs — cancel via tasks/cancel before the status flips to running for a free abort.

​
Poll

curl https://llm.bytespike.ai/v1/tasks/query?task_id=task_01AbCdEf \
  -H "x-api-key: $BYTESPIKE_API_KEY"
Response when complete: 
{
  "task_id": "task_01AbCdEf",
  "status": "completed",
  "result": {
    "video_url": "https://cdn.bytespike.ai/vid/...",
    "duration_seconds": 5,
    "resolution": "1080p"
  },
  "credits": 0.42
}
Recommended polling cadence: 2-second intervals up to 180 seconds, then back off to 5 seconds. The query endpoint is free.

​
When to use

  • Need 1080p output → Sora 2 or Sora 2 Pro
  • Need vertical (9:16) for short-form social → Sora 2 supports it natively
  • Latency-sensitive → consider Veo 3.1 Fast or Seedance Fast for ~30s turn-around at lower fidelity