generated: '2026-08-17'
method: searched
source: >-
  https://api.cloud.quandela.com/openapi.json (4xx response descriptions on
  POST /api/jobs and the /qt/* Quantum Toolbox operations), plus live
  anonymous header inspection of https://api.cloud.quandela.com/health and
  /api/status on 2026-08-17
note: >-
  Quandela publishes NO conventional request-rate limit — there is no
  requests-per-second/minute figure anywhere in the spec or the docs, no 429
  status is declared on any of the 59 operations, and no RateLimit-*,
  X-RateLimit-* or Retry-After response header is defined in the spec or
  returned on live responses. What Quandela DOES publish, inside the OpenAPI
  4xx response descriptions, is a set of real CONCURRENCY and RESOURCE
  ceilings on quantum job admission. Those are captured below because they are
  the limits an agent will actually hit. Every ceiling is published as a
  template placeholder ({MAX_WAITING_JOBS}, {shots_limit}) rather than a
  number, except max_duration, so the concrete value is only knowable from the
  error response at runtime.
limit_count: 6
response_headers:
  ratelimit_standard: null
  x_ratelimit: null
  retry_after: null
  observed_on:
    - url: https://api.cloud.quandela.com/health
      status: 200
      headers_present: [Server, Date, Content-Type, Content-Length, Connection]
      rate_limit_headers: none
  note: >-
    No rate-limit signalling of any kind on the wire. An agent has no
    runtime budget signal and must treat the ceilings below as discoverable
    only by failing.
exhaustion_status_codes:
  - status: 400
    used_for: waiting-job ceiling, max_duration ceiling, max_shots ceiling
  - status: 403
    used_for: concurrent Quantum Toolbox job ceiling
  - status: 401
    used_for: credit / platform-time exhaustion (NOT an auth failure — see errors/quandela-problem-types.yml)
  - status: 429
    used_for: null
    present: false
limits:
  - name: waiting_jobs
    scope: per-account (global or offer-specific)
    window: concurrent
    limit: '{MAX_WAITING_JOBS}'
    limit_published_as: placeholder
    signal: 400 "Cannot create more than {MAX_WAITING_JOBS} jobs"
    operation: openapi/quandela-perceval-job-openapi.yml#post_api_jobs
  - name: running_quantum_toolbox_jobs
    scope: per-account
    window: concurrent
    limit: '{MAX_QT_JOBS}'
    limit_published_as: placeholder
    signal: 403 "Cannot have more than {MAX_QT_JOBS} running Quantum Toolbox jobs"
    operation: openapi/quandela-quantum-toolbox-openapi.yml#post_qt_chemistryvqe
  - name: max_duration_absolute
    scope: per-job
    window: per-request
    limit: 864000
    unit: seconds
    limit_published_as: concrete
    signal: 400 "Max duration too large, exceeding 864000"
    operation: openapi/quandela-perceval-job-openapi.yml#post_api_jobs
  - name: max_duration_per_platform
    scope: per-platform
    window: per-request
    limit: '{limit}'
    unit: seconds
    limit_published_as: placeholder
    signal: 400 "Max_duration {value}(s) exceeds the platform limit of {limit}(s)"
    operation: openapi/quandela-perceval-job-openapi.yml#post_api_jobs
  - name: max_shots_per_platform
    scope: per-platform
    window: per-request
    limit: '{shots_limit}'
    limit_published_as: placeholder
    signal: 400 "Max shots must less than or equal to {shots_limit}"
    operation: openapi/quandela-perceval-job-openapi.yml#post_api_jobs
  - name: credit_balance
    scope: per-account
    window: prepaid-balance
    limit: account credit balance (and, on Enterprise offers, platform time)
    limit_published_as: account-specific
    signal: >-
      401 "Not enough credits" / "Not enough credits: all remaining credits are
      reserved by waiting or running jobs" / "Not enough {platform_type} time left"
    operation: openapi/quandela-perceval-job-openapi.yml#post_api_jobs
adjacent_capacity_signals:
  - name: platform availability
    url: https://api.cloud.quandela.com/api/platforms/public
    status: 200
    auth: none
    note: >-
      Anonymous endpoint returning per-QPU `status`, a `maintenance`
      {start_time, end_time} window, and a `statistic` map of daily
      availability percentages for the trailing ~8 days. Not a rate limit, but
      it is the only capacity signal Quandela publishes machine-readably, and
      the correct pre-flight check before submitting a job.
  - name: job availability
    operation: openapi/quandela-perceval-job-openapi.yml#get_api_jobs_availability
    auth: job token
    schema: JobAvailabilityResponse
    fields:
      [max_concurrent_jobs, num_concurrent_jobs, max_jobs_in_queue,
       num_jobs_in_queue, max_running_qt_jobs, num_running_qt_jobs]
    note: >-
      The one place the ceilings become concrete. This authenticated endpoint
      returns the account's actual max_concurrent_jobs, max_jobs_in_queue and
      max_running_qt_jobs alongside current usage — i.e. the numbers the
      {MAX_WAITING_JOBS} and {MAX_QT_JOBS} placeholders stand in for. Because
      the values are per-offer and behind a job token, they are not publishable
      here; an agent should call this before submitting rather than relying on
      the spec text. Effectively a body-carried substitute for RateLimit-*
      headers, but it must be polled explicitly rather than arriving with every
      response.
  - name: token usage report
    operation: openapi/quandela-perceval-job-token-openapi.yml#get_api_tokens_usage_report
    auth: bearer
    note: Per-token usage statistics with start_time/end_time query window — the consumption ledger.
gaps:
  - No published requests-per-interval limit.
  - No 429 Too Many Requests anywhere in the spec.
  - No RateLimit-* / X-RateLimit-* / Retry-After headers on any response.
  - Concrete ceiling values are placeholders in the spec; discoverable only from a failed request.