Rate limits & budgets

Each organization has a requests-per-minute limit and an optional spend cap. Both are enforced at the gateway and shown live in your console.

Requests per minute

Your org has a per-minute request limit (RPM). Exceeding it returns 429 with a Retry-After header. The exact number is per-org configuration — see it live under Usage & budget in the console (the Rate limit metric).

Budget cap

An org can have a maximum spend (max_budget_nano). As requests are billed, spend accrues against this cap. When the cap is reached, further requests are blocked (429) until the budget is raised or reset. Current spend, the cap, and a usage bar are shown in the console.

Staying under the limits

• Batch or queue bursts so you stay within your RPM.
• Implement exponential backoff on 429, honoring Retry-After.
• Watch spend vs. budget in the console analytics and raise the cap before you hit it.