generated: '2026-07-20'
method: searched
source: https://docs.modulate.ai/guides/authentication
docs: https://docs.modulate.ai/guides/authentication
summary: >-
  Rate limits are enforced per organization, per model. Two primary limits apply:
  a concurrency limit (maximum simultaneous in-flight requests or active WebSocket
  connections) and a monthly usage quota (maximum audio hours processed per calendar
  month). An optional organization-wide monthly credit (spend) cap can also be set,
  disabled by default. New accounts start at 1-5 concurrent streams per model,
  self-service adjustable up to 25; higher limits via support@modulate.ai.
limits:
- name: concurrency
  scope: per-organization-per-model
  description: Maximum simultaneous in-flight requests or active WebSocket connections.
  default_min: 1
  default_max: 5
  self_service_ceiling: 25
- name: monthly-usage
  scope: per-organization-per-model
  unit: audio-hours
  description: Maximum audio hours processed per calendar month, set by Modulate per model.
- name: monthly-credit
  scope: per-organization
  description: Optional organization-wide monthly spend cap. Disabled by default.
responses:
- transport: rest
  status: 403
  meaning: Monthly usage quota exceeded, or model access not enabled.
- transport: rest
  status: 429
  meaning: Too many concurrent requests (concurrency limit hit).
- transport: websocket
  close_code: 4029
  meaning: Rate limit exceeded — monthly quota or concurrency limit hit at handshake.
retry_guidance: >-
  Implement exponential backoff with jitter on 429. Concurrency limits are per-model,
  so distributing load across models can help avoid hitting them. Monthly quota
  exhaustion requires waiting for the period reset or contacting support.
headers_documented: false
usage_dashboard: https://platform.modulate.ai/dashboard/usage