Skip to main content

Quotas & Rate Limits

The gateway enforces two independent controls before forwarding a request upstream, both evaluated during the authentication step: a per-Gateway-Key requests/minute quota, and a per-Environment IP allowlist.

Per-key requests/minute quota​

Each Gateway Key optionally carries a requests-per-minute limit (unset = unlimited, configured at key creation). When set, every request increments a rolling 60-second counter for that key. If the increment would exceed the configured limit, the request is rejected before it reaches the provider:

429 Gateway Key request limit exceeded

This check happens after the key itself is resolved but before the IP allowlist check, so an invalid key still fails with 401 rather than 429.

IP allowlist​

Each Environment can be configured with an IP allowlist. If the list is non-empty and the caller's IP isn't in it, the gateway rejects the request:

403 IP address not allowed for this environment

An empty allowlist means no IP restriction.

Where this fits in the request lifecycle​

Provider Resolution (URL)
  → Gateway Key Resolution
    → Request-Limit-Per-Minute Quota (optional)
      → IP Allowlist Check
        → Provider Key Resolution
          → Forward to Provider

Both checks are cheap and run before any call to the upstream provider - a rejected request never reaches (and never bills against) the provider.