token_quota_exceeded — key token quota exhausted

The order was not created. This is the only 429 with a detailed body: window (the first window violated), limit, used, requested, reset_at, retry_after. It comes together with a Retry-After header.

The quota counts tokens, not requests — the accounting unit for the volume of work done by the service (this is NOT money and NOT LMC). The per_minute / per_hour / per_day / per_month windows come from the limits.tokens.* of your key. The check happens BEFORE creation: on POST /orders — against the estimate, on POST /orders/{order}/retry — against the sum of the original chunks, on regenerate — against the new text.

Wait until reset_at — or retry_after seconds; it is the same value. The presence of Retry-After is exactly how you tell this case apart from the key rate limit, which has no such header at all. If you cannot wait, reduce requested: split the text into parts so that each one fits into what remains in the window (limit − used).

The same request once Retry-After has elapsed. The active windows of the key are visible in the developer section (limits.tokens.*); a key without token limits is not limited by volume at all.

• Guide: limits and the 429 code
• 429 — all five sources
• limit_exceeded — subscription limit, not key limit
• POST /orders
• My key limits