Please clarify the last update

Hi.
Changelog says that rate limits were changed (increased) Changelog

can you please tell me, to what numbers were increased? lets say, what are for candles & trades.

Hello @ttodua! Candles and Trades endpoints are both in the Public Endpoints category. See the changes in rate limits below.

Before:

Rest API

  • For Public Endpoints, our rate limit is 3 requests per second, up to 6 requests per second in bursts.
  • For Private Endpoints, our rate limit is 5 requests per second, up to 10 requests per second in bursts.

FIX API

  • The FIX API rate limit is 50 messages per second.

After:

Rest API

  • For Public Endpoints, our rate limit is 10 requests per second, up to 15 requests per second in bursts.
  • For Private Endpoints, our rate limit is 15 requests per second, up to 30 requests per second in bursts.

FIX API

  • The FIX API rate limit is 50 commands per second, and up to 100 messages per second in bursts.

Please note that some endpoints may have a custom rate limit like the /fills endpoint which has a custom rate limit of 10 requests per second, and up to 20 requests per second in bursts.You may check the latest information about rate limits on our documentation at Rate limits.

If you have further questions, feel free to reply to this thread. Thanks!

2 Likes

Thanks for reply!
Can you please tell, what is explanation of your term “burst”. a regular RL is 10/s, but 15/s in bursts, please give me an example to find out how many times we can request 15/s continiously.

Hello @ttodua, the burst limit is the number of requests our API can handle concurrently. As to how rate limits work, you may refer to our documentation site as it gives an example with a formula: Rate limits

Thank you!

2 Likes

the doc page you linked says:

TokenBucket with burst = 3 and refresh_rate = 1

However, that information is not clear. for example, public endpoints have: 10 requests per second, up to 15 requests per second in bursts. so, that translates to :

10-regular | 15 burst ---- refresh-rate 1 second

however, that doc page just says:

TokenBucket with burst = 3 and refresh_rate = 1

and it misses the info “regular” detail (how much is “regular” rate-limit" for that example? as it only mentions “burst” rate limit).
do you understand my confusion? thanks

Hi @ttodua, apologies for the delay in our response.

The refresh rate would match the regular rate limit (eg: for Exchange public endpoints the token bucket will be filled with 10 tokens per second). So the token bucket can hold a maximum of 15 tokens, allowing a “burst” of 15 requests in a second. But only 10 tokens get added back in the next second, so if 15 requests occur again in the very next second, there will only be 10 tokens to satisfy the request and 5 of the requests will be denied with a response code of 429.

Although Public and Private endpoints have slightly different rate limits, the example below works for both in principle.

  • Public endpoints have 10 requests per second, and 15 requests per second in burst.
  • An easy way to think about it is everyone starts at the burst rate (in this case they can send 15 requests per second).
  • Each request we receive decrements their outstanding requests per second limit by 1.
  • We credit their limits at a rate which results in 10 requests per second (so for an example you can imagine we credit them 1 request every 0.1 seconds).
  • If they don’t send requests frequently enough to consume the 10 requests per second, after enough time has passed with us crediting them 1 request every 0.1 seconds, they will eventually top their request limits bucket up to the burst rate which is the max (15 in this case).

Hope this helps! Thank you!

2 Likes

@shirley.dizon - your answer has been really exceptional! thank you for such good explanation, it just answered all the details and questions I had, perfect in-detail description! It’s worth that it was included in docs :slight_smile:

4 Likes