# Provenance stamped 2026-10-07 (roadmap#943): written by an API Evangelist sweep (undated — the file carries no date and git cannot supply one), not
# harvested from the provider. `method: generated` so the economics campaign may replace it with a
# dated, verbatim transcription; every figure below is unverified until it does.
method: generated
rateLimits:
  - api: Akash Console Deployment API
    baseURL: https://console-api.akash.network
    notes: >-
      Formal rate limit thresholds not publicly published. API returns HTTP 429
      TooManyRequests when the rate is exceeded. Recommended best practices from
      documentation: avoid polling more frequently than every 3 seconds; use
      exponential backoff on 5xx responses.
    limits:
      - type: advisory
        requests: 1
        unit: per 3 seconds (recommended polling interval)
        enforcement: HTTP 429 returned when exceeded
        retry: exponential backoff recommended

  - api: Akash Console Network Data API
    baseURL: https://console-api.akash.network
    notes: >-
      Public, unauthenticated API. No formal rate limits documented in official
      Akash documentation as of June 2026. Reasonable use expected; avoid
      aggressive scraping.
    limits:
      - type: none-documented
        notes: No formal rate limits published for public endpoints

  - api: Akash Blockchain REST API
    baseURL: https://api.akashnet.net
    notes: >-
      Node operators are advised to implement rate limiting via reverse proxy
      at approximately 10 requests per second for production deployments.
      Public community nodes (rpc.akashnet.net, polkachu, etc.) may impose
      their own limits independently.
    limits:
      - type: recommended
        requests: 10
        unit: per second
        enforcement: Reverse proxy (e.g., nginx) enforced by node operator
        notes: Applied per-IP at proxy layer; not enforced by the Akash protocol itself

  - api: AkashML Inference API
    baseURL: https://api.akashml.com
    notes: >-
      No formal rate limits published as of June 2026. Service advertises
      auto-scaling with no capacity caps and sub-200ms response times across
      80+ global datacenters. Contact support for enterprise volume needs.
    limits:
      - type: none-documented
        notes: No formal rate limits published; auto-scaling infrastructure claimed