model sume/auto on /v1/videos vs pinning Seedance 2.5: what differs

Use sume/auto when you want a clip and do not care which model makes it; pin seedance-2.5 when you need its 4 to 30 second range, a specific resolution grid or a price you can compute in advance. The docs say the Auto create controls default to 720p and 8 s, with 3 to 10 s clips at 16:9 or 9:16, and that Sume never discloses which family served the request (Video generation).

Side by side

Auto request

The request omits resolution and lets Sume apply its defaults. A replay with the same Idempotency-Key gets the same route and price.

curl -X POST https://api.sume.com/v1/videos \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: auto-001" \
  -d '{
    "model": "sume/auto",
    "prompt": "A vertical UGC-style product clip on a desk, natural light",
    "aspect_ratio": "9:16",
    "duration": 5
  }'

Which to use

• Exploration and short clips: sume/auto.
• 15 to 30 seconds in one job: pin seedance-2.5.
• Repeatable per-model budgets: pin an id, because you can compute its price.
• Do not infer the served family from the output; Sume says to avoid that.

Pricing a pinned run

With a pinned id you can price the job before sending it. An 8-second 720p Seedance 2.5 clip at 9:16 is $4.6224, and a 30-second one is $17.334 at 720p. With sume/auto the route and price are resolved from the normalized request and the catalog version, so a replay with the same idempotency key lands on the same route and the same price; the first request is the only place the price is unknown to you in advance.

The older /v1/video-router/* endpoints still work and create the same jobs, but the docs recommend sume/auto on POST /v1/videos for new integrations.