Billing and limits FAQ

Are requests inside my plan allowance charged per token?

No. Plan-included usage carries no per-token charge; per-token prices apply only to usage paid from credits.

Why did I get 402 quota_exhausted?

The plan allowance is used up and credits cannot cover the request. The limit field says which cap applied; see quota_exhausted.

How do I stop one team from spending the whole budget?

Give each team a workspace with its own spend limit; teams can request increases that an admin approves.

Is the batch discount automatic?

Yes: requests submitted through Batches are billed at the batch rate. The same request sent normally is billed at the standard rate.

Do retries cost money?

Each request that produces output is billed, including retries. Requests refused before running (for example capacity_busy, usage_check_unavailable) are not.

Which costs more: extended context or splitting the document?

Extended context is a preview design and its pricing is Proposed — not offered (no long-context rate is set). Retrieval or chunking already keeps each request small when a question only needs part of the material; see Extended context.