specification: API Commons Rate Limits
specificationVersion: '0.1'
provider: Nace.AI — NDI
providerId: nace-ai
generated: '2026-07-20'
method: searched
created: '2026-07-20'
modified: '2026-07-20'
tags:
  - Rate Limiting
  - Document Intelligence
description: >-
  NDI enforces per-tenant default rate limits documented on the Errors & limits
  page. Requests-per-minute and sync-wait concurrency throttle to 429 (with a
  Retry-After header) or degrade to 202 polling; credit exhaustion surfaces as
  402 quota_exceeded (not retryable).
sources:
  - https://docs.ndi.nace.ai/concepts/errors
headers:
  requestId: X-Request-Id
  retryAfter: Retry-After
responseCodes:
  throttled: 429
  quotaExceeded: 402
limits:
  - name: Requests per minute (per tenant)
    scope: tenant
    metric: requests_per_minute
    limit: 60
    timeFrame: minute
    on_exceed: 429 too_many_requests with Retry-After header; respect it then retry
  - name: Concurrent jobs (pending + running)
    scope: tenant
    metric: concurrent_jobs
    limit: 10
    timeFrame: concurrent
    on_exceed: 429 too_many_requests
  - name: Concurrent sync waits
    scope: tenant
    metric: concurrent_sync_waits
    limit: 4
    timeFrame: concurrent
    on_exceed: degrades to 202 (poll GET /api/v1/jobs/{job_id})
credits:
  included: 50000
  on_exhaustion: 402 quota_exceeded (not retryable; raise credit limit)
notes: >-
  Values are documented defaults and may be raised per account. GET
  /api/v1/health/load returns per-client in-flight and windowed counts so
  callers can self-throttle before hitting the limits.