Runway Model Routers report the model used; Sume Auto does not

If you let a router pick the video model, can you see which model ran and what it cost? On Runway, yes: its pricing page says router generations are billed at the standard rate of whichever model the router selects, and the response metadata reports the model used and the realized cost in credits. On Sume, model: "sume/auto" is the opposite by design: the docs say the response echoes sume/auto and the family that ran is never disclosed.

What does Runway document for routers?

Runway Dev sells credits at $0.01 each, and every generation costs credits. The router sentence on its pricing page is short: routed generations are billed at the standard rate of the selected model, and the metadata names the model and the realized cost in credits. That gives you an audit trail per generation. For scale, the same page lists wan3 at 5, 10 and 20 credits per second at 480p, 720p and 1080p, which is $0.05, $0.10 and $0.20 per second at the $0.01 credit price (arithmetic).

What does Sume Auto document?

sume/auto is a Sume-only value for model on POST /v1/videos. The docs say Sume picks the family, the poll response reports "model": "sume/auto", and you should not build on any observable trait of the output to infer which family served the request. Resolution is a pure function of the normalized request and the catalog version, so an idempotent replay prices and routes the same way.

Auto create controls default to 720p and 8 seconds, with 3 to 10 second clips at 16:9 or 9:16. Billing is the usual video rule: the workspace USD balance is reserved at submit at provider list times 1.25, and usage.cost is the Sume billable amount.

What does that mean for cost control?

With a router that reports the model, you can decompose spend after the fact. With Auto, the control you have is the request itself: the duration, resolution and aspect ratio you send, and the reserve the API takes against your balance. If you need a known price per second, pin a catalog model instead of using Auto. The Video Router page documents per-model limits such as seedance-2.5 at 4 to 30 seconds and wan-3.0 at 2 to 30 seconds, and says to read capabilities from GET /v1/video-router/models rather than assuming one envelope.

What should you log either way?

Log your own request fields, the job id, the final cost, and for Sume the pinned model id when you pin one. If your reporting needs the model name per clip, plan around that now: with Auto it is not available to you, so the choice is to pin. Neither page says which approach gives better video, so test a few prompts on each before committing a budget.