Open API rate limits

About this document

This document defines rate limiting rules for the Open API, including limits by dimension and time window, so developers can control call frequency and avoid failures from exceeding limits.

Core definitions

1. Rate limit dimensions

ID	Name	Description
USER	User	Limits per user identifier (e.g. user ID)
APP	Application	Limits per application identifier (e.g. app ID linked to API key)
GLOBAL	Global	Platform-wide limit across all callers

2. Time windows

ID	Unit	Description
SECOND	Second	1-second window
MINUTE	Minute	1-minute (60-second) window

Rate limit rules

The platform uses multi-dimensional, multi-window limits. Exceeding any dimension/window triggers throttling:

Dimension	Window	Limit	Description
USER	Second	100/sec	Max 100 calls per second per user
USER	Minute	5000/min	Max 5000 calls per minute per user
APP	Second	20/sec	Max 20 calls per second per app
APP	Minute	1000/min	Max 1000 calls per minute per app
GLOBAL	Second	3000/sec	Max 3000 calls per second platform-wide
GLOBAL	Minute	150000/min	Max 150000 calls per minute platform-wide

Limits follow a "lowest ceiling wins" rule. For example, 101 calls in 1 second triggers throttling even if the minute limit is not reached; 1001 calls in 1 minute triggers throttling even if the per-second limit is not reached.

When limits are hit

1. Status code: The API returns 60001 (INTERFACE_RATE_LIMIT):
   • Code: 60001
   • Message: Rate limit exceeded, please try again later
2. Recovery: Wait for the current window to end (1 second for second-level limits, remaining minute time for minute-level limits), then counts reset.

Notes

1. Counts are based on requests received and processed; invalid parameters and auth failures still count toward limits.
2. Implement client-side throttling and split bursts into smaller batches per these rules.
3. Contact platform operations if you need higher limits.
4. Rules may change as capacity evolves; changes will be announced in advance.