Rate Limits
Key-gated endpoints are rate-limited per plan, with a per-minute and (below Enterprise) a per-month budget. The budget is pooled across every API key on the account. Extra keys don't add throughput. Every response carries quota headers, and exceeding a limit returns 429 with Retry-After.
Per-plan limits
Limits are pooled per account (per subscription), not per key: every active key on the account draws from one shared budget, so minting more keys never multiplies your throughput. Use extra keys for isolation and rotation (e.g. one per environment), not capacity. You can hold up to 5 active keys per subscription; creating a 6th returns 400 until you revoke one. The minute window smooths bursts; the monthly window caps total volume (Enterprise has no monthly cap).
Quota headers
Every key-gated response includes the minute-window headers, so you can throttle proactively instead of waiting for a 429. Plans with a monthly cap (Developer, Pro) also include the -Month headers; Enterprise has no monthly cap and so omits them:
Handling 429
When you exceed either window the API returns 429 Too Many Requests with a Retry-After header (seconds to wait). Back off for that long, then retry, preferably with exponential backoff and jitter for bursts. Cache where you can: the NBBO and pricing feeds are meant to be polled at a steady cadence, not hammered.
Market Health
Market health, the one keyless endpoint, is limited separately by IP (roughly 60 requests / 60s) and returns no per-key quota headers. Every other endpoint is key-gated; see Authentication to get a key.