Concurrent seats

A seat holds one stream of synthesis at a time — a live conversation, a narration job, a batch of renders — and everything said on it is unmetered. A metered plan sells the seat and then bills per character on top of it. Here the seat is the whole bill, so the only number you size is how much you need running at once.

1 seat

one stream at a time — usage on it unmetered

The quiet cap in metered plans

The shape is the same everywhere: you buy a seat, and parallel capacity means buying more seats. What differs is what happens on top. On a per-character plan you also pay for every character, and a mid-tier plan caps how many seats you may buy at all — past that ceiling the next step is a sales call. Here you buy the same seats, usage on each is unmetered, and there is no cap that turns into a sales call.

The three mechanics

Capacity, priced three ways

Fig

1 seat

one stream at a time, everything on it unmetered — $150 a month annual

$10

a burst seat-day, when traffic spills past your committed seats

$20 / mo

a standby seat idling beside another vendor, for failover

A request that cannot be served is refused fast and retryably — one busy lane never quietly slows the others.

Sizing a fleet

Count busy-hour simultaneous conversations, not monthly minutes — that peak is your seat count.
Let bursts spill instead of sizing for them: the spike days cost $10 a seat-day, capped in the dashboard, and the quiet days cost nothing extra.
The pilot exists to find the real number: 20 seats against 30 days of actual traffic.