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: Adept
providerId: adept
created: '2026-05-08'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-08, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-08'
reconciled: true
tags:
- AI
- Agents
- Foundation Models
- Action Models
- Open Source
- Rate Limiting
- Quotas
- Throttling
description: >-
  Adept does not operate a public commercial API and therefore does not
  publish API-level rate limits. Practical rate limits for users of Adept's
  open-source models (Fuyu-8B, Persimmon-8B) are determined by the
  customer's own self-hosted inference stack or by their chosen third-party
  inference provider, not by Adept.
notes: >-
  No commercial API exists to throttle. Hugging Face's normal download
  rate limits apply when pulling model weights.
sources:
- https://www.adept.ai/
- https://huggingface.co/adept
responseCodes:
  throttled: 429
limits:
- name: Commercial API
  scope: not_applicable
  metric: not_offered
  limit: not offered
  notes: Adept does not operate a commercial inference API to rate-limit.
- name: Hugging Face Downloads
  scope: huggingface_account
  metric: downloads
  limit: per Hugging Face Hub policy
  notes: Standard Hugging Face anonymous / authenticated download limits apply when fetching weights.
- name: Self-Hosted Inference
  scope: deployment
  metric: requests
  limit: bounded by self-hosted GPU capacity
  notes: Throughput is determined by the customer's chosen hardware.
policies:
- name: Self-Host Capacity Planning
  description: Size GPU deployments per Fuyu-8B / Persimmon-8B model footprint and expected concurrency.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com