generated: '2026-08-13'
method: searched
source: https://docs.aws.amazon.com/clean-rooms/latest/userguide/clean-rooms-quotas.html
docs: https://docs.aws.amazon.com/clean-rooms/latest/userguide/quotas.html
console: https://console.aws.amazon.com/servicequotas/home/services/cleanrooms/quotas
limit_count: 6
note: >-
  AWS Clean Rooms publishes a per-operation request-rate quota for every API
  action through AWS Service Quotas. The overwhelming default is 5 requests per
  second per Region per account; two poll-heavy read operations get 20/sec. Every
  rate quota is adjustable via Service Quotas. Because the surface is one quota
  per operation (roughly 100 rows), the entries below record the DISTINCT rate
  tiers plus the resource limits an agent has to plan around, and point at the
  full published table rather than mirroring 100 identical rows.
signalling:
  response_headers: []
  header_note: >-
    AWS Clean Rooms returns no RateLimit-* / X-RateLimit-* / Retry-After headers.
    There is no runtime budget signal on a successful response — the only
    exhaustion signal is the error itself, which is a meaningful agent-readiness
    gap: a client cannot see how close it is to the ceiling before it hits it.
  exhaustion_status: 400
  exhaustion_error: ThrottlingException
  exhaustion_message: >-
    "Your request rate is too high. The AWS SDKs automatically retry requests
    that receive this exception. Reduce the frequency of requests."
  retry_strategy: >-
    Client-side exponential backoff with jitter, implemented automatically by
    every AWS SDK (adaptive/standard retry mode). No server-supplied retry hint.
  source: https://docs.aws.amazon.com/clean-rooms/latest/apireference/CommonErrors.html
rate_limits:
  - id: default-api-rate
    scope: per-account-per-region-per-operation
    window: 1s
    limit: 5
    unit: requests
    adjustable: true
    applies_to: >-
      The default for essentially every cleanrooms API action — Create*, Delete*,
      Update*, Get*, List*, BatchGet*, StartProtectedQuery, StartProtectedJob,
      PopulateIdMappingTable, PreviewPrivacyImpact, TagResource, UntagResource.
  - id: get-configured-table-rate
    scope: per-account-per-region-per-operation
    window: 1s
    limit: 20
    unit: requests
    adjustable: true
    applies_to: GetConfiguredTable
  - id: get-protected-query-rate
    scope: per-account-per-region-per-operation
    window: 1s
    limit: 20
    unit: requests
    adjustable: true
    applies_to: GetProtectedQuery
    note: >-
      The higher ceiling here matters: GetProtectedQuery is the polling call an
      agent uses to await an asynchronous protected query result.
  - id: concurrent-sql-queries
    scope: per-account-per-region
    window: concurrent
    limit: 5
    unit: running SQL queries
    adjustable: true
    note: 'us-east-1: 5; each other supported Region: 2.'
  - id: concurrent-pyspark-jobs
    scope: per-account-per-region
    window: concurrent
    limit: 5
    unit: running PySpark jobs
    adjustable: true
    note: 'us-east-1: 5; each other supported Region: 2.'
  - id: concurrent-queries-per-membership
    scope: per-membership
    window: concurrent
    limit: 5
    unit: ongoing queries
    adjustable: false
resource_limits:
  - name: Query run time
    limit: 24 hours
    note: Maximum duration a query runs before timeout.
  - name: Query text length
    limit: 90 KB
  - name: Query text length (differential privacy)
    limit: 90 KB
    note: >-
      The quotas table also lists an 8 KB figure for differential-privacy query
      text; the resource-parameter table and the 2025-07-16 doc-history entry
      both state 90 KB. Recorded as published; the two tables disagree.
  - name: Query text length on Spark analytics engine
    limit: 500 KB
  - name: Collaborations created per account
    limit: 500
    adjustable: true
  - name: Memberships per account
    limit: 500
    adjustable: true
  - name: Configured tables per account
    limit: 250
    adjustable: true
  - name: Configured tables per protected query
    limit: 15
    adjustable: true
  - name: Members invited per collaboration
    limit: 5
    adjustable: true
  - name: Analysis rule size
    limit: 100 KB
    adjustable: false
  - name: Columns per configured table allowlist
    limit: 100
    adjustable: true