generated: '2026-09-02'
method: searched
source: https://developer.unico.io/developers/api-reference/rate-limits
docs: https://developer.unico.io/developers/api-reference/rate-limits
limit_count: 2
rate_limits:
- id: tenant-default-rps
  scope: per-tenant
  window: 1s
  limit: 10
  unit: requests
  burst: null
  configurable: true
  note: >-
    "The default value is 10 RPS per tenant." Exact limits are configured per tenant; Unico
    directs integrators to their Onboarding team to confirm the applied ceiling and to request
    an increase BEFORE volume rises.
  source: https://developer.unico.io/developers/api-reference/rate-limits
- id: webhook-concurrency
  scope: per-webhook-endpoint
  window: concurrent
  limit: 500
  unit: in-flight deliveries
  burst: null
  configurable: true
  note: >-
    Maximum simultaneous in-flight webhook deliveries, set when the webhook is registered.
    Beyond it Unico queues deliveries. Distinct from the request budget: this one throttles
    Unico calling the integrator.
  source: https://developer.unico.io/developers/webhooks-and-events/security
response_headers:
- header: Retry-After
  on_status: 429
  units: seconds
  required_behavior: honor it
  observed: documented
  example: 'Retry-After: 12'
- header: X-RateLimit-Limit
  present: false
- header: X-RateLimit-Remaining
  present: false
- header: X-RateLimit-Reset
  present: false
- header: RateLimit
  present: false
exhaustion:
  status_code: 429
  reason_phrase: Too Many Requests
  fallback_when_header_absent: >-
    Exponential backoff 1s, 2s, 4s, 8s ... capped at 60s.
concurrency:
  note: >-
    Separate from the per-second budget, the platform enforces infrastructure-level
    concurrency caps. Unico's guidance is to target a steady RPS rather than bursting the
    whole minute's budget into one second.
adaptive_throttling:
  applies_to: webhook delivery
  behavior: >-
    Sustained non-2xx responses from an integrator's webhook endpoint cause Unico to
    automatically reduce throughput to that endpoint. Returning 503 during planned maintenance
    triggers it; Unico advises stopping intake at the load balancer instead.
  source: https://developer.unico.io/developers/webhooks-and-events/security
polling_guidance:
  note: >-
    Unico explicitly names polling GET /client/v1/process/{id} as the fastest way to exhaust
    the tenant budget and directs integrators to subscribe to the PROCESS_STATE_FINISHED
    webhook instead. Token minting per request is called out the same way — cache the
    access token for its full 1-hour TTL.
observed_live: false
observed_note: >-
  Every documented endpoint requires an authenticated Bearer token, so no unauthenticated
  live response was available to read headers from. Values above are documented, not probed.