specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Stream
providerId: getstream
created: '2026-07-03'
modified: '2026-07-03'
reconciled: false
tags:
- Chat
- Rate Limiting
- Quotas
- Throttling
description: >-
  Stream applies rate limits per API endpoint and per platform on a rolling
  1-minute window. Platforms (Server-side, iOS, Android, Web) have independent
  counters and do not share quota, so a server-side script hitting a limit does
  not affect mobile or web clients. Each response returns the current quota in
  the X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset headers,
  and a 429 status is returned when a limit is exceeded. Per-user endpoints are
  additionally capped at roughly 60 requests per minute per endpoint and
  platform; app-level endpoint limits range from about 60 to 10,000 requests per
  minute. Applications can query their live quotas via GET /rate_limits. WebSocket
  connections are separately capped at 50 concurrent connections per user in
  production (3 in development).
notes: >-
  Numeric limits vary per endpoint and may be adjusted dynamically based on
  overall platform load and an application's usage patterns for query endpoints.
  Monitor the X-RateLimit-* response headers and call GET /rate_limits to see
  current per-platform, per-endpoint quotas.
sources:
- https://getstream.io/chat/docs/node/rate_limits/
- https://getstream.github.io/protocol/?urls.primaryName=Chat
- https://getstream.io/chat/docs/node/instantiating_the_client/
responseCodes:
  throttled: 429
headers:
- name: X-RateLimit-Limit
  description: Total requests allowed in the current window for this endpoint and platform.
- name: X-RateLimit-Remaining
  description: Requests remaining in the current window.
- name: X-RateLimit-Reset
  description: Unix timestamp when the current window resets.
limits:
- name: Per-User Endpoint Limit
  scope: user
  metric: requests
  limit: ~60 per minute per endpoint per platform
  window: 1 minute
  notes: Each individual user is capped per endpoint and platform.
- name: Per-App Endpoint Limit
  scope: app
  metric: requests
  limit: ~60 to 10,000 per minute (varies by endpoint)
  window: 1 minute
  notes: Application-wide limit that varies by endpoint; not shared across platforms.
- name: Platform Isolation
  scope: platform
  metric: requests
  limit: independent counters per platform
  window: 1 minute
  notes: Server-side, iOS, Android, and Web each have their own quota pool.
- name: Concurrent WebSocket Connections
  scope: user
  metric: connections
  limit: 50 in production, 3 in development
  window: concurrent
  notes: connectUser() should be called once per user session; extra sockets cause billing and behavior issues.
policies:
- name: Per-Endpoint Per-Platform Windows
  description: Limits are enforced on a 1-minute rolling window, isolated by endpoint and platform.
- name: Dynamic Adjustment
  description: Query-endpoint limits may be raised or lowered based on platform load and the application's usage patterns.
- name: Backoff Strategy
  description: Clients should honor the X-RateLimit-Reset timestamp and implement exponential backoff with jitter on 429 responses.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com