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 Quantum AI
providerId: google-quantum-ai
created: '2026-05-25'
modified: '2026-05-25'
reconciled: true
tags:
  - Quantum Computing
  - Quantum Engine
  - Quotas
  - Reservations
description: |
  The Quantum Engine API itself does not publish per-second request rate limits in the same way as
  consumer Google APIs. Throughput is governed by:
    - Standard Google Cloud per-project quotas applied via the Cloud Quotas system.
    - Per-job runtime caps (5 minutes maximum during OPEN_SWIM time slots).
    - Reservation budgets that allocate total processor minutes/hours to a project.
    - Processor TimeSlot scheduling (OPEN_SWIM, MAINTENANCE, RESERVATION, UNALLOCATED).
  Approved users share processor capacity; queued jobs are scheduled by the Engine based on
  reservation priority and OPEN_SWIM availability.
sources:
  - https://quantumai.google/cirq/google/concepts
  - https://quantumai.google/cirq/google/access
  - https://quantumai.google/cirq/google/engine
headers:
  limit: x-goog-quota-user
  remaining: ''
  reset: ''
  retryAfter: retry-after
responseCodes:
  throttled: 429
  quotaExceeded: 429
algorithm: reservation-and-time-slot
quotas:
  - name: Per-job runtime (open swim)
    scope: job
    limit: 300
    unit: seconds
    description: A single job may not occupy more than 5 minutes of processor time during an OPEN_SWIM
      slot.
  - name: Processor reservation window
    scope: reservation
    limit: variable
    unit: minutes
    description: Granted as part of a reservation grant; consumed by jobs targeted at the reservation's
      processor during its window.
  - name: Cloud Quotas — list/get/create requests
    scope: project
    limit: variable
    unit: requests
    description: Standard Google Cloud per-minute API quotas applied to QuantumEngineService methods
      via the Cloud Quotas system. Visible and adjustable in the GCP console.
backoff:
  strategy: exponential
  initialDelay: 1s
  maxDelay: 60s
  jitter: true
  description: Cirq's engine client retries transient gRPC UNAVAILABLE / RESOURCE_EXHAUSTED errors with
    exponential backoff; user code should mirror this pattern when calling the REST surface directly.
slotTypes:
  - OPEN_SWIM
  - MAINTENANCE
  - RESERVATION
  - UNALLOCATED