30-second AI video generator: which models make it in one request

Two models in Sume's catalog generate a 30-second video in one request: seedance-2.5 (4 to 30 seconds) and wan-3.0 (2 to 30 seconds). The other text and image-to-video models top out at 15 seconds; only the Motion Transfer model, which is listed when its provider is configured, also accepts up to 30. Source: the video docs.

ByteDance's launch post also states up to 30 seconds per generation for Seedance 2.5.

What does a 30-second clip cost?

Reserved amounts for 30 seconds, 16:9, from Sume's pricing code.

How do I request it?

Set duration to 30 and send the resolution and ratio explicitly.

curl -X POST https://api.sume.com/v1/videos \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.5",
    "prompt": "A day in a fishing village, dawn to dusk, unhurried cuts",
    "duration": 30,
    "resolution": "720p",
    "aspect_ratio": "16:9"
  }'

Should I use one 30-second take or stitched clips?

It depends on the ad or scene. One request keeps lighting and identity consistent; stitched shots let you retry a single scene. See make a 30-second commercial with AI for the shot-by-shot route.