specification: API Commons Rate Limits
specificationVersion: '0.1'
provider: Langdock
providerId: langdock
generated: '2026-07-19'
method: searched
created: '2026-07-19'
modified: '2026-07-19'
tags:
- Rate Limiting
- Artificial Intelligence
description: >-
  Langdock enforces default per-model rate limits on the public API expressed as both
  a request-rate and a token-rate. Workspaces that bring their own model keys (BYOK)
  can have admins configure custom per-model limits in Settings.
sources:
- https://docs.langdock.com/en/developer/overview/api-introduction
responseCodes:
  throttled: 429
limits:
- name: Requests per minute (default, per model)
  scope: workspace
  metric: requests_per_minute
  limit: 500
  timeFrame: minute
- name: Tokens per minute (default, per model)
  scope: workspace
  metric: tokens_per_minute
  limit: 60000
  timeFrame: minute
custom_limits:
  supported: true
  applies_to: BYOK workspaces
  detail: Admins can configure custom limits per model in Settings.
notes: >-
  The 429 "Rate limit exceeded" response is declared on 11 operations in the OpenAPI
  (openapi/langdock-openapi-original.yml), including the completion, embedding, agent,
  skills, and audit-log endpoints. The spec does not declare RateLimit-* response
  headers, so no header contract is recorded here.