Runway model routers vs sume/auto: what you can steer

Sume's sume/auto takes no cost, latency or quality preference: you give a prompt, optional ratio and duration, and Sume chooses the family without telling you which. Runway's models page says its Model Routers automatically select the best model based on cost, latency or quality preferences at no extra charge (Runway API models, read 2026-10-10).

That makes them different tools: Runway's is a preference-driven router, Sume's is a default that you use when you do not care which family runs.

What sume/auto does

Per the Video Generation docs, model: "sume/auto" is an addition only Sume has. The poll response reports sume/auto and Sume does not disclose the family that ran. Resolution is a pure function of the normalized request and the catalog version, so an idempotent replay gets the same price and the same route. The Video Router page says Auto controls default to 720p and 8 seconds, with 3 to 10 second clips at 16:9 or 9:16.

What you can and cannot steer

Side by side:

When to pin instead

Pin an id when output has to be repeatable across a campaign, when you need more than 10 seconds, 1:1 or other ratios, or an input type the Auto pipe does not offer. Record the id with each clip, because a different row changes the price and the look.

• Test copy ideas with sume/auto; it costs the same as a pinned row and needs no migration when a model launches.
• Pin wan-3.0 or seedance-2.5 for lengths up to 30 seconds.
• Pin gemini-omni-flash-1.1 for edits, 4K or reference clips.

One caution on the legacy route

The Video Router's generate route does not accept routing_preset; it takes a catalog id or an Auto alias. If an older script passes a preset, move it to a pinned id or to /v1/videos with sume/auto.