Openai Sora 2 Text To Video API Documentation

Openai Sora 2 Text To Video API Documentation

Playground

OpenAI Sora 2 is a state-of-the-art text-to-video model with realistic visuals, accurate physics, synchronized audio, and strong steerability. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

> Notice — Service Stability > > The Sora 2 family is being retired or has already been retired by its original provider. Until it is fully removed, the service is unstable: generations may fall back to alternative models without notice, results can differ from what this model used to produce, and the service can be temporarily unavailable. > > If you need an equally capable, stable alternative, we recommend Seedance 2: bytedance/seedance-2.0/text-to-video.

Sora 2 Text-to-Video

Describe the scene in natural language, reference your pre-defined character IDs, and the model renders a cohesive, temporally consistent video where every character looks and moves exactly as intended — no manual compositing required.

Why Choose This?

• True multi-character consistency Reference two or more character IDs in a single generation. Each character retains its unique appearance, proportions, and style throughout every frame.

• Natural-language scene control Describe interactions, environments, and actions in plain text. The model understands spatial relationships and character dynamics to produce believable compositions.

• Flexible aspect ratio support Choose between portrait (720×1280) and landscape (1280×720) orientations to match your target platform.

• Scalable duration Generate clips from 4 seconds up to 20 seconds in fixed steps, giving you full control over pacing and output cost.

• Production-ready output Delivers smooth, artifact-free motion suitable for marketing content, storytelling, game cinematics, and social media video.

Parameters

How to Use

1.2. Select size — portrait (720×1280) for mobile/social, landscape (1280×720) for widescreen. 3. Set duration — choose 4, 8, 12, 16, or 20 seconds based on your scene length. 4.5. Submit — generate, preview, and download your video.

Pricing

Billing Rules

• Rate: $0.10 per second
• Duration options: 4, 8, 12, 16, or 20 seconds
• Billing is based on the selected duration, not actual playback length

Best Use Cases

• Social Media Content — Produce portrait-format multi-character clips optimized for Reels, TikTok, and Shorts.
• Advertising & Campaigns — Generate diverse cast scenarios rapidly for A/B testing creative variations.

Pro Tips

• Be specific about character positions and actions in your prompt for better spatial composition.
• Use portrait mode (720×1280) for mobile-first platforms and landscape (1280×720) for cinematic or desktop use.
• Start with a 4-second generation to validate composition and character rendering before committing to a longer duration.
• Ensure all referenced character IDs are valid and accessible in your account before submitting.

Notes

• Character IDs must be created and saved in advance — this model references existing character profiles and does not create new definitions.
• Complex multi-character scenes benefit from concise, clearly structured prompts.

Related Models

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "size": "720*1280",
  "duration": 4
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/openai/sora-2/text-to-video" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Every model also accepts the global request options enable_sync_mode and enable_base64_output; see their pages for details.

Response Parameters

Result Request Parameters

Result Response Parameters