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: Esri
providerId: esri
created: '2026-05-04'
modified: '2026-09-07'
generated: '2026-09-07'
method: searched
source: >-
  https://doc.arcgis.com/en/arcgis-online/manage-data/limit-usage-of-secure-services.htm,
  https://developers.arcgis.com/rest/geoenrichment/service-limits/,
  https://developers.arcgis.com/rest/geoenrichment/error-codes/,
  https://location.arcgis.com/pricing/
supersedes: >-
  The 2026-05-04 bulk-sweep scaffold, which asserted X-RateLimit-Limit / X-RateLimit-Remaining /
  X-RateLimit-Reset / RateLimit-Policy response headers and three invented per-minute tiers.
  Esri publishes none of those headers and none of those tiers. Replaced with what is actually
  documented, including the honest gaps.
tags:
  - Geographic
  - Geospatial
  - GIS
  - Location
  - Mapping
  - Maps
  - Spatial Analysis
  - Rate Limiting
  - Quotas
  - Throttling
description: >-
  Esri does not publish a numeric per-key request rate for ArcGIS Location Services. What is
  documented is (a) a per-organization query rate ceiling on ArcGIS Online with a one-minute
  cooling-off period and HTTP 429, (b) hard per-operation size limits on GeoEnrichment and
  routing, and (c) monthly usage allowances that are metering, not throttling. No rate-limit
  response headers are documented on any surface.
limit_count: 6
headers:
  limit: null
  remaining: null
  reset: null
  retryAfter: null
  policy: null
headers_note: >-
  NO rate-limit response headers are documented for ArcGIS Location Services or the ArcGIS REST
  API. Not X-RateLimit-*, not RateLimit-*, and no Retry-After on 429. An agent has no runtime
  signal to back off against and must use a fixed 60-second wait after a 429, because that is
  the documented cooling-off window. This is a real and material gap for agent use.
responseCodes:
  throttled: 429
  quotaExceeded: 429
  authFailure: 401
  authFailureBodyCodes: [498, 499]
  serviceUnavailable: 503
body_error_caution: >-
  ArcGIS also signals failures inside a 200 body as {"error":{"code":...}}. Do not treat a 200
  as success without parsing.
limits:
  - name: ArcGIS Online organization query rate
    scope: per-organization
    metric: queries_per_minute
    limit: 10000
    burst: null
    timeFrame: minute
    tier: standard
    source: https://doc.arcgis.com/en/arcgis-online/manage-data/limit-usage-of-secure-services.htm
    note: >-
      Standard organizations share a 10K/minute query ceiling; premium data stores scale up to
      96K/minute (M4). Exceeding it triggers a one-minute cooling-off period during which
      non-repeatable queries return HTTP 429.
  - name: ArcGIS Online organization query rate (premium data store, M4)
    scope: per-organization
    metric: queries_per_minute
    limit: 96000
    timeFrame: minute
    tier: premium
    source: https://doc.arcgis.com/en/arcgis-online/manage-data/limit-usage-of-secure-services.htm
  - name: Cooling-off period after throttle
    scope: per-organization
    metric: seconds
    limit: 60
    timeFrame: incident
    note: >-
      A one-minute cooling-off window is triggered when the rate ceiling is hit. No Retry-After
      header is sent; 60 seconds is the documented wait.
  - name: GeoEnrichment study areas per request
    scope: per-request
    metric: study_areas
    limit: 1000
    source: https://developers.arcgis.com/rest/geoenrichment/error-codes/
    error_code: 10040001
  - name: GeoEnrichment maximum ring buffer
    scope: per-request
    metric: miles
    limit: 1000
    source: https://developers.arcgis.com/rest/geoenrichment/error-codes/
    error_code: 10040004
  - name: GeoEnrichment service area ceiling
    scope: per-request
    metric: composite
    limit: '300 miles drive distance / 300 minutes drive time / 27 miles walking distance / 540 walking minutes'
    source: https://developers.arcgis.com/rest/geoenrichment/error-codes/
    error_code: 10040006
service_limits_discovery:
  endpoint: https://geoenrich.arcgis.com/arcgis/rest/services/World/geoenrichmentserver/Geoenrichment/serviceLimits?f=pjson
  note: >-
    GeoEnrichment exposes its own machine-readable limits via a serviceLimits discover method —
    a genuinely good pattern, and the only self-describing limits surface Esri publishes.
usage_allowances:
  note: >-
    The monthly free allowances on location.arcgis.com/pricing (2M basemap tiles, 20K geocodes
    not stored, 20K routes, 50K elevation values, 5K service areas, 1K static maps, 1K basemap
    sessions, 500 places searches) are BILLING thresholds, not rate limits. Crossing them
    charges the account; it does not throttle it. See plans/esri-plans-pricing.yml.
policies:
  - name: Always send a token
    description: >-
      Esri documents that rate limiting is applied to free operations and that including a token
      with every request — even a free one — raises the applicable limit and reduces throttling.
      Anonymous calls are the ones that get slowed down.
  - name: Backoff
    description: >-
      With no Retry-After and no RateLimit-* headers, use a fixed 60-second wait on 429 followed
      by exponential backoff with jitter.
  - name: Retries are billable
    description: >-
      Successful location-service calls are metered. Retrying a call that actually succeeded
      (200 with an error body misread, or a client timeout after server success) bills twice.
undocumented:
  - Per-API-key request rate for ArcGIS Location Services (geocode-api, route-api, places-api, elevation-api)
  - Concurrent-request ceilings
  - Any rate-limit response header on any surface
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com