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: Advanced Micro Devices
providerId: advanced-micro-devices
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
notes: AMD does not publish per-call rate limits for Developer Cloud or ROCm. Capacity is
  governed by allocated GPU hours and concurrent-job limits, not HTTP throttling.
tags:
  - Rate Limiting
  - GPU
  - HPC
description: AMD's developer surface is GPU-allocation-based rather than request-rate-based; no
  public per-second or per-minute rate-limit numbers are documented.
sources:
  - https://www.amd.com/en/developer.html
  - https://rocm.docs.amd.com/
responseCodes:
  throttled: 429
  serviceUnavailable: 503
limits:
  - name: Developer Cloud allocation
    scope: account
    metric: varies
    limit: governed by allocated GPU hours and concurrent jobs
policies:
  - name: Allocation-scoped
    description: Capacity governed by per-account GPU-hour and concurrency allocation rather than
      request throttling.
  - name: Backoff
    description: Honor Retry-After when present on any HTTP API surface.