Contact
Questions about models, volume, limits, or how a particular workload would be billed — all of it is a conversation with the people who run the service, not a ticket in a queue.
A public support address is being set up, and we would rather say so than publish one nobody reads yet.
Most of what people ask is already answerable here without it. The API reference covers the endpoints, the limits and every error code; pricing covers what a package includes and how usage is billed; and the questions on the front page cover the rest.
If something is still open after that, create an account — it takes a moment, it needs no card, and your keys are ready the moment you sign in.
Tell us the models you plan to run and roughly how much traffic, and we will come back with a rate and a sensible starting balance. Moving an existing workload? The models, the average request and response sizes, and whether you stream are the useful things to include.
Adding tokens, raising a rate limit, or lifting a concurrency ceiling for a batch job. Mention the timing if it is a one-off burst, and it is set before the burst starts rather than during it.
If something you need is not in the catalogue, ask. Availability changes, and the list on this site is generated from what is actually callable right now.
Create one first — it takes a moment, it needs no card, and it is the address we will reply to. Your keys are ready the moment you sign in, so there is nothing to wait for on our side.