10-second AI video API: Sume models that accept 10 seconds and cost

Ten seconds is inside the range of every text-to-video model in the Sume catalog: Wan 3.0 (2 to 30), Seedance 2.5 (4 to 30), Seedance 2 and its Fast and Mini variants (4 to 15), Kling 3 (4 to 15), MiniMax H3 and H3 Max (5 to 15) and Gemini Omni Flash 1.1 (3 to 10, its ceiling). The cheapest per-second row is Wan 3.0 at 480p, which is $0.63 for 10 seconds.

Price for 10 seconds

Sume bills the provider list times 1.25, rounded up to the cent per job. Seedance models are priced per 1,000 video tokens, so read their pricing_skus in the listing instead of a per-second figure.

Which ones do not take a duration

Genjutsu and H3 Max Recast transform a source video and use its length, and Grok Imagine Video 1.5 needs a source image. Send those with the inputs they require, and no text-only prompt.

The request

Send an Idempotency-Key so a retry after a timeout reuses the same job.

curl -X POST "https://api.sume.com/v1/videos" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "minimax-h3",
    "prompt": "A drone shot over terraced rice fields",
    "duration": 10,
    "resolution": "768p",
    "aspect_ratio": "16:9"
  }'

Wait for it properly

A 10-second render is not instant. Poll the polling_url, or set callback_url to an HTTPS endpoint and verify the webhook signature, rather than holding a connection open.