Rate limits

Each API key has up to four budgets, configurable per key in the dashboard:

• rate_limit_per_second — token bucket, optionally with a higher burst_limit for short spikes.
• rate_limit_per_minute, rate_limit_per_hour, rate_limit_per_day — fixed windows.

All configured windows are enforced server-side. When you exceed any of them, the request returns 429 rate_limited with a Retry-After header (in seconds) and a detail naming which window was hit.

Recommended client behavior

• On 429, wait Retry-After seconds before retrying. Don't hammer.
• If you expect bursts (e.g. batch jobs), ask for a burst_limit on the key rather than retry-looping.
• Async endpoints (?async=true) consume a single rate-limit slot at enqueue time, not per poll.