Sora API — Text-to-Video in an OpenAI-Compatible Workflow

Sora is OpenAI's text-to-video model, and “the Sora API” is how teams generate video programmatically rather than through the consumer app. On Kunavo today, text-to-video generation runs on Google Veo 3 through a single OpenAI-style video endpoint — Sora access is on the roadmap, and because the endpoint is model-agnostic, migrating to Sora later will be a one-word change. This guide shows the Veo 3 workflow so every example works now.

What is the Sora API?

Sora (Sora 2 and Sora 2 Pro) turns a text prompt — or a still image — into a short video clip, with synchronized audio in Sora 2. The API approach lets you script generation within a pipeline: marketing shots, product animations, b-roll, and storyboard previews. The format is the same across modern video models: you send a prompt and parameters and receive the URL of a hosted video in return.

Video on Kunavo today: Veo 3

Kunavo exposes video generation through a single OpenAI-style endpoint, /v1/video/generations. Sora is not yet enabled in the catalog; the live text-to-video model is Google Veo 3, which produces cinematic clips with native audio. The examples below use veo-3 — when Sora arrives, the only change will be the model field.

Text-to-video quickstart

A POST request with your Kunavo key. Generation can take several minutes, so the synchronous call keeps the connection open until the clip is ready:

import requests

resp = requests.post(
    "https://api.kunavo.com/v1/video/generations",
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={
        "model": "veo-3",
        "prompt": "um dolly-in cinematográfico em um origami de garça vermelha se abrindo, luz suave",
        "aspect_ratio": "16:9",
    },
    timeout=600,  # a geração pode levar minutos
)
print(resp.json()["data"][0]["url"])

duration (seconds) sets the clip length on the models billed per second: the Seedance models and Wan 2.7. The Veo models ignore it; every Veo clip is 8 seconds, billed per video.

Image-to-video

To animate a still image, pass image_url (an https URL or a file uploaded to /v1/files) together with the prompt. For controlled motion, you can pass a first and last frame with image_urls and image_mode: "frame". Complete examples are available in the video documentation.

Asynchronous task lifecycle

In production, do not hold a connection open for 10 minutes. Submit a task to /v1/videos, receive a task ID immediately, and then poll GET /v1/videos/{id} until it completes. Result URLs are permanent.

# Produção: envie uma tarefa e faça polling — sem conexão de longa duração.
task = requests.post(
    "https://api.kunavo.com/v1/videos",
    headers={
        "Authorization": f"Bearer {API_KEY}",
        "Idempotency-Key": "meu-uuid-de-tarefa",   # seguro para retry em ~24h
    },
    json={"model": "veo-3", "prompt": "...", "aspect_ratio": "16:9"},
    timeout=60,
).json()
# depois faça polling em GET /v1/videos/{task["id"]} até concluir

The video documentation covers the complete polling loop, idempotency keys, and webhook delivery.

Prices

Veo 3 is charged per video (for an 8-second 720p clip, shown here), about 50–70% below Google's list price. Higher resolutions cost more — see the pricing page for the complete tier table.

Video prompt tips

• Describe the shot, not just the subject. Camera movement (dolly, pan, push-in), lens feel, lighting, and pacing matter more than adjectives.
• Define the aspect ratio explicitly — 16:9 for landscape, 9:16 for vertical/social media.
• Every Veo clip is 8 seconds long. Plan each shot to fit, and stitch multiple generations for longer sequences.
• Use a reference frame (image-to-video) when you need a specific character or product to remain consistent.

Frequently asked questions