Each API key has a per-minute request limit. Over the limit, out of plan allowance or out of balance, a request is rejected at once and not charged.
You can also set a spend limit and a token limit for each key on the API keys page.
Key points
- Limits apply per API key and active plan.
- Each request is charged only once, even when it is retried internally.
- Dashboard figures can lag a few seconds behind the actual limits.
- When streaming, cap the total data your app accepts.