Pin a video model id or use sume/auto when new models keep launching?
Pin an id like seedance-2.5 for repeatable batches; use sume/auto for 3-10 s drafts. What each setting guarantees and what Auto hides, per Sume's docs.
Pin a catalog id such as seedance-2.5 when a batch must be repeatable or needs a clip length Auto does not offer; use sume/auto when you want Sume to pick for a 3-10 second clip. A pinned job records the id you asked for, while Auto aliases stay opaque. As new models launch, a pinned id changes only when you change it.
Rules come from Sume's Video Router docs, Video generation docs and API reference, read 2026-10-02.
What does pinning guarantee?
job.model equals the routable id you requested. You also get that model's full envelope: seedance-2.5 and wan-3.0 take up to 30 seconds, minimax-h3 takes 5-15 s, and gemini-omni-flash-1.1 takes 3-10 s with 4K. Billing is provider list times 1.25 on every row.
A pinned id also stays routable after Sume stops choosing it by default: the docs say seedance-2 and seedance-2-mini are no longer picked by any default or routing preset but stay routable when requested.
| Setting | Model recorded on the job | Clip length | Defaults |
|---|---|---|---|
model: "seedance-2.5" | seedance-2.5 | 4-30 s | Model defaults; set resolution explicitly |
model: "sume/auto" | Opaque Auto alias | 3-10 s at 16:9 or 9:16 | 720p and 8 s |
What does Auto give up?
Auto create controls are limited to 3-10 second clips at 16:9 or 9:16, so it cannot make a 20-second clip. Sume's public responses are provider-neutral and Auto aliases stay opaque, which means you cannot tell from the job which model ran, so Auto is a poor fit when you must reproduce a look.
What it gives in return is not choosing at all, which suits a first draft.
How should I decide when a new model arrives?
Check whether the new id is in the catalog first, using the one-call catalog check. If it is, test it pinned at 480p against your current id, as in same prompt on three models, then switch the id in one place.
Keep Auto for exploration and pinned ids for anything a customer will see twice. Store the id in config, so a switch is one edit.
If you are unsure, start with a pinned id in production and Auto in your internal tools. The pinned id gives you a stable look and a stable bill per clip, and Auto lets people explore without learning the catalog.
Sources
Related posts
More in Developers
- Pin a TTS model id: sonic-latest or sonic-3.6 on Sume
Voice vendors move aliases under you. How sonic-latest and sonic-3.6 behave on the Sume TTS router, and why to send an explicit id in production.
- PNG, JPEG or WebP for ad stills: measure file size with Pillow
Sume's Image API has no output_compression field. Request png, jpeg or webp, then re-encode locally and compare sizes before you upload.
- A policy snapshot file and a Python check before each AI ad batch
Keep platform AI rules in a JSON file with source and read date, and run a short Python check that flags rules older than 30 days before you queue a batch.
- Poll a Sume bulk-run queue in Python and list items not completed
A short asyncio script that polls GET /v1/format-run-queues/{id} until the queue is completed, prints the counts and lists every item that did not complete.
Written by Sume