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: Haystack / deepset
providerId: haystack-ai
created: '2026-06-20'
modified: '2026-06-20'
reconciled: false
tags:
- AI
- LLM
- RAG
- Open Source
- Rate Limiting
- Quotas
- Throttling
description: >-
  The open-source Haystack framework and self-hosted Hayhooks deployments are not
  rate limited by deepset - throughput is bounded only by your own infrastructure
  and by the model/embedding providers your pipelines call. The hosted deepset
  Cloud REST API applies request timeouts (about 2 minutes for search and 3
  minutes for other requests) and standard pagination, and enforces
  account/plan-level quotas that are governed by enterprise agreement rather than
  a published per-minute table. Specific numeric limits are not reconciled in
  this artifact.
notes: >-
  Confirm hosted deepset Cloud quotas and any per-account limits with deepset
  during reconciliation; values are governed by the customer's plan and contract.
  Self-hosted Haystack/Hayhooks have no provider-imposed limits.
sources:
- https://docs.cloud.deepset.ai/reference/api-overview
- https://docs.cloud.deepset.ai/docs/api/
- https://github.com/deepset-ai/haystack
responseCodes:
  throttled: 429
limits:
- name: Search Request Timeout
  scope: request
  metric: duration
  limit: ~120 seconds
  notes: Search requests time out at roughly 2 minutes.
- name: Other Request Timeout
  scope: request
  metric: duration
  limit: ~180 seconds
  notes: Non-search requests time out at roughly 3 minutes.
- name: Pagination Limit
  scope: request
  metric: items
  limit: see provider documentation
  notes: Page-based (page_number/limit) and cursor-based (before/after) pagination.
- name: Account Quotas
  scope: account
  metric: requests
  limit: governed by plan/contract
  notes: Hosted deepset Cloud quotas are set by enterprise agreement, not published.
- name: Self-Hosted (Haystack / Hayhooks)
  scope: deployment
  metric: requests
  limit: none imposed by provider
  notes: Throughput bounded only by your infrastructure and downstream model providers.
policies:
- name: Plan-Governed Limits
  description: Hosted limits are defined per customer plan and enterprise agreement.
- name: Backoff Strategy
  description: Clients should implement exponential backoff with jitter and honor Retry-After on 429.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com