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: Intel
providerId: intel
created: '2026-05-04'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-04, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-05'
reconciled: false
tags:
  - Confidential Computing
  - Attestation
  - oneAPI
  - Rate Limiting
description: Intel does not publish per-second rate-limit numbers in the public Intel Trust Authority
  REST API documentation. Limits are operationally enforced as fair-use throttling on shared infrastructure.
  Intel oneAPI portals are documentation / download surfaces rather than runtime APIs and do not have
  per-call rate-limit semantics worth capturing here.
notes: Numeric per-second limits are not published; recorded as 'fair-use throttling'. Customers requiring
  guaranteed throughput should use the Intel Trust Authority paid-support tier.
sources:
  - https://docs.trustauthority.intel.com/main/articles/restapi-intro.html
responseCodes:
  throttled: 429
  serviceUnavailable: 503
limits:
  - name: Intel Trust Authority attestation throughput
    scope: api-key
    metric: varies
    limit: fair-use throttling on shared attestation infrastructure
    applies:
      - Intel Trust Authority API
  - name: Regional availability
    scope: region
    metric: varies
    limit: US/World endpoint and EU endpoint; unavailable in mainland China
    applies:
      - Intel Trust Authority API
policies:
  - name: Backoff Strategy
    description: Clients receiving 429 or 503 should implement exponential backoff with jitter and honor
      Retry-After when present.
  - name: Region Selection
    description: Use the EU endpoint (https://api.eu.trustauthority.intel.com) for EU-resident workloads
      and the global endpoint (https://api.trustauthority.intel.com) elsewhere; the service is unavailable
      in mainland China.
  - name: Token Caching
    description: Cache issued attestation tokens for the duration of their validity rather than re-attesting
      on every workload start to reduce request volume.
  - name: Fair Use
    description: Intel reserves the right to throttle traffic that materially impacts shared infrastructure;
      paid support customers can request guaranteed-throughput SLAs.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com