Skip to main content
The quota depends on your plan. One request = 1 unit, regardless of how many rows are returned. The counter resets at the top of each hour (UTC).

​
Plan access

Some endpoints are reserved for higher plans: Calling an endpoint outside your plan returns 403:
403

​
Quota headers

Every authenticated response carries:

​
Over quota (429)

Once the quota is exhausted, requests return 429 until the reset:
429
The Retry-After header gives the number of seconds until the reset (top of the next hour, UTC).

​
Burst limit: 50 requests / second

On top of the hourly quota, each key is capped at 50 requests per second. This limit protects service availability: without it, a single client could burn its hourly quota in seconds and saturate the API for everyone. Beyond that, the response is 429 — without consuming quota:
429
The Retry-After header is 1: the counter resets on the next second, unlike the hourly quota which waits for the top of the hour. In practice, spread your calls rather than sending bursts — 50 requests per second is far more than enough to feed a real-time model.

​
One WebSocket connection per key

The WebSocket feed (/ws, Sharp plan) accepts one simultaneous connection per API key. Every socket receives the full feed: multiple connections on the same key multiply bandwidth without adding anything. A second connection is refused with close code 4409:
The new connection is the one refused, never the existing one: an established feed will not be dropped in favour of a duplicate — typically a script restarted without closing the previous one. If your client disconnected abruptly, the slot frees up when the server closes the socket, and reconnecting immediately works. Need several parallel feeds? Create one key per process from your account area.

​
Track your usage

Your current-hour usage is also shown in your account area.