Skip to main content
Rate limits meter requests by IP, API key, or key and IP. Exhaustion returns 429 rate_limited.

​
Retry timing

IP limiters send X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset, but no Retry-After. Key limiters send these headers only on denial, with Retry-After in seconds. Reset is an absolute Unix timestamp in milliseconds. Wait for Retry-After when present; otherwise use the reset timestamp.

​
IP defaults

Limits below are per minute. Orchestration names refer to /v1/orchestration/, except the v2 route catalog.
  • v2 routes and limits: 60 requests/IP.
  • Unauthenticated estimate and authenticated order lookup: 120 requests/IP.
  • SSE connection attempts, affiliate claims, and attestation: 30 requests/IP.
Status requires authentication; it has no anonymous polling mode.

​
Key defaults

Client keys use two buckets. Values below are requests per minute, per key and per key+IP respectively. Partner overrides can change defaults and apply key limits to server keys.
  • Quote: 600 / 60.
  • Served estimate and status: 6000 / 1200.
  • Submit and onramp: 120 / 10.
  • Accumulation create/reindex and liquidation create: 60 / 5.
Authenticated estimates use the served-estimate bucket; legacy pricing can also charge the quote-budget limits. Poll no faster than every 3 seconds per order. Use webhooks or SSE for updates; contact support for higher quotas.