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: Google Vault
providerId: google-vault
generated: '2026-09-12'
method: searched
source: https://developers.google.com/workspace/vault/limits
created: '2026-05-04'
modified: '2026-09-12'
tags:
  - Rate Limiting
  - Quotas
  - Throttling
description: >-
  Google publishes the Vault API's quotas in full, as per-minute per-project request budgets
  split by resource family, plus one organization-wide ceiling and one product limit on
  concurrent exports. It publishes NO rate-limit response headers, so an agent cannot read its
  remaining budget at runtime — the only runtime signal is the 429 itself.
note: >-
  Replaces the 2026-05-04 scaffold, which asserted invented free-tier numbers and
  X-RateLimit-* headers that Google does not send. Every figure below is verbatim from the
  provider's usage-limits page, read on 2026-09-12.
headers:
  limit: null
  remaining: null
  reset: null
  retryAfter: null
  policy: null
headers_note: >-
  None published. Google documents no X-RateLimit-*, no RateLimit-* and no Retry-After for
  Vault. Quota consumption is visible only in the Google Cloud console quota pages, out of band.
responseCodes:
  throttled: 429
  quotaExceeded: 429
retry:
  strategy: truncated exponential backoff
  formula: min(((2^n)+random_number_milliseconds), maximum_backoff)
  maximum_backoff_seconds: [32, 64]
  jitter: random_number_milliseconds, <= 1000 ms, recalculated per retry
  source: https://developers.google.com/workspace/vault/limits
product_limits:
  - name: Concurrent exports
    scope: organization
    limit: 20
    unit: exports in progress
    quote: You can have no more than 20 exports in progress across your organization.
limits:
  - name: Matter reads, organization-wide
    scope: organization
    metric: matter_reads
    limit: 600
    timeFrame: minute
    applies: [Vault API, vault.google.com]
    note: >-
      600 matter reads per minute across all projects and users, and it counts traffic from the
      Vault web UI as well as the API — an agent shares this ceiling with the humans in the
      organisation.
  - name: Export, matter and saved query reads
    scope: project
    metric: read_requests
    limit: 120
    timeFrame: minute
  - name: Hold reads
    scope: project
    metric: read_requests
    limit: 228
    timeFrame: minute
  - name: Long-running operation reads
    scope: project
    metric: read_requests
    limit: 300
    timeFrame: minute
  - name: Export writes
    scope: project
    metric: write_requests
    limit: 20
    timeFrame: minute
  - name: Hold writes
    scope: project
    metric: write_requests
    limit: 60
    timeFrame: minute
  - name: Matter permission writes
    scope: project
    metric: write_requests
    limit: 30
    timeFrame: minute
  - name: Matter writes
    scope: project
    metric: write_requests
    limit: 60
    timeFrame: minute
  - name: Saved query writes
    scope: project
    metric: write_requests
    limit: 45
    timeFrame: minute
  - name: Search counts
    scope: project
    metric: count_requests
    limit: 20
    timeFrame: minute
quota_cost_by_method:
  note: >-
    A single call can spend several budgets at once, and some calls cost more than one unit.
    matters.list is the trap — one call spends TEN matter reads against a 600/minute
    organization-wide ceiling, so naive polling of the matter list exhausts the organisation's
    budget roughly ten times faster than the request count suggests.
  methods:
    - methods: [matters.close, matters.create, matters.delete, matters.reopen, matters.update, matters.undelete]
      costs: [1 matter read, 1 matter write]
    - methods: [matters.count]
      costs: [1 count]
    - methods: [matters.get]
      costs: [1 matter read]
    - methods: [matters.list]
      costs: [10 matter reads]
    - methods: [matters.addPermissions, matters.removePermissions]
      costs: [1 matter read, 1 matter write, 1 matter permissions write]
    - methods: [matters.exports.create]
      costs: [1 export read, 10 export writes]
    - methods: [matters.exports.delete]
      costs: [1 export write]
    - methods: [matters.exports.get]
      costs: [1 export read]
    - methods: [matters.exports.list]
      costs: [5 export reads]
    - methods: [matters.holds.addHeldAccounts, matters.holds.create, matters.holds.delete, matters.holds.removeHeldAccounts, matters.holds.update]
      costs: [1 matter read, 1 matter write, 1 hold read, 1 hold write]
    - methods: [matters.holds.list]
      costs: [1 matter read, 3 hold reads]
    - methods: [matters.holds.accounts.create, matters.holds.accounts.delete, matters.holds.accounts.list]
      costs: [1 matter read, 1 matter write, 1 hold read, 1 hold write]
    - methods: [matters.savedQueries.create, matters.savedQueries.delete]
      costs: [1 matter read, 1 matter write, 1 saved query read, 1 saved query write]
    - methods: [matters.savedQueries.get]
      costs: [1 matter read, 1 saved query read]
    - methods: [matters.savedQueries.list]
      costs: [1 matter read, 3 saved query reads]
    - methods: [operations.get]
      costs: [1 long-running operation read]
quota_increase:
  available: true
  note: >-
    Google states a project can request a quota adjustment, that calls by a service account
    count as a single account, and that approval is not guaranteed.
  source: https://developers.google.com/workspace/vault/limits
limit_count: 10