Rate limits

During the alpha, each API key is limited to 2 requests per second and each source IP is limited to 20 requests per second across keys. Partner and Creator API surfaces have separate IP buckets. Successful responses describe the per-key budget:

Header	Example	Meaning
`RateLimit`	`limit=2, remaining=1, reset=1`	Limit, remaining requests, and seconds until reset.
`RateLimit-Policy`	`2;w=1`	Two requests per one-second window.

A limited request returns 429 and a Retry-After header:

{ "error": "Too many requests, please try again later." }

Wait for Retry-After, add jitter when many workers share a key, and avoid concurrent retry loops. The IP cap can be reached even if every individual key remains below its own limit.

Overview

Core concepts