Rate limits
How many calls a key and an address may make, and what happens past the limit.
Each key may make 600 reads and 120 changes a minute, and each address 1,200 calls a minute, whatever key it sends. A read is a GET or HEAD; a change is anything else.
Every answer carries three headers:
| Header | Meaning |
|---|---|
X-RateLimit-Limit |
The calls of this kind the key may make in a minute, or the address's limit when no key was sent. |
X-RateLimit-Remaining |
How many of them are left in this minute. |
X-RateLimit-Reset |
When the minute ends, in Unix seconds. |
Over the limit, the answer is 429 with the code rate_limited and a Retry-After header saying how many seconds to wait. Wait that long before trying again, rather than retrying at once.
The limits are checked before the key is, so a flood of calls with a wrong key is refused just as quickly.
Keeping inside the limits
- Give each integration a key of its own, so a busy one cannot use up another's allowance.
- Read a list once and work from it, rather than reading each item in turn. People come up to 200 a page with
?limit=200. - Use webhooks to hear about a change instead of asking for it again and again.