Aggregator: one subscription, several calendars
An aggregator sells access to models from several makers. One entry in this register names six, which means six retirement schedules and authority over none of them. Nothing on its page describes how such a decision would be passed on to a subscriber. As of 2026-09-22.
| For the subscriber | For the aggregator |
|---|---|
| One bill and one interface | Several suppliers, each with its own terms |
| Several makers' schedules, unseen | No authority over any retirement |
| A working product when one line goes | A menu that has to be edited |
| A broken series, if it used that line | No published route for passing the news on |
Inclusion rule. Distinctions drawn from the entries in this register that name models from more than one maker on their own pages. Order. The subscriber's position first, then the seller's.
1Aggregation helps the product and not the production
Losing one of six models leaves a functioning subscription and a broken series, if the series was established on the line that went. The consolation is real for the customer as a customer and worthless to them as a producer.
That distinction is why this register tracks model lines rather than product features. A feature list says what an app can do this month; a model list says whose roadmap somebody has quietly signed up to.
2Menus lag, and that is normal
A model list is a marketing page, so it changes when somebody updates it rather than when a maker publishes a date. One entry here still names a model whose maker has announced its interface closes.
Both statements are current and the register keeps them side by side. Resolving the tension would mean predicting what the application does next.
3Where the same thing happens at the maker layer
Four entries here make their own models and resell others on the same plan table. That is aggregation with a floor under it: if every licensed row went, they would still have something to sell.
The page setting the two layers against each other shows what difference the floor makes and what it does not.
A definition rather than a dated entry: no line above is a vendor statement. Where this word turns up in a real notice, the wording and the date are on the invideo AI reading, which names six makers. Nearby terms: Credit conversion, Output unit, Tier gating.
4A pipeline across model versions
Model end dates touch “Aggregator”; SceneMixer's guides describe how its pipeline treats the models it runs:
Doubao Seed 2.1 Pro is the family’s full-size tier at 100% of the base text rate, twice the default. Same three jobs — reading, writing, storyboards — with more headroom on long, crowded episodes: it holds the storyboard’s strict format more reliably and keeps track of many characters and locations in one pass.
SceneMixer, Doubao Seed 2.1 Pro (read 2026-10-13)
Wan 3.0 · 480P is the tier every new project starts on.
SceneMixer, Wan 3.0 · 480P: the default video tier (read 2026-10-13)