generated: '2026-08-17'
method: searched
source: https://docs.pyannote.ai/ratelimits
docs: https://docs.pyannote.ai/ratelimits
limit_count: 3
window_note: Every limit uses a fixed 60-second window and is scoped per team, per endpoint.
limits:
- scope: per-team-per-endpoint
  applies_to:
  - POST /v1/diarize
  - POST /v1/identify
  - POST /v1/voiceprint
  - POST /v1/live
  - POST /v1/media/input
  - POST /v1/media/output
  limit: 100
  unit: requests
  window: 60s
  description: >-
    Default limit for submitting jobs, creating streams and the media endpoints. Each endpoint has its
    own independent 100 req/min budget.
- scope: per-team-per-endpoint
  applies_to:
  - GET /v1/jobs/{jobId}
  - GET /v2/jobs
  limit: 300
  unit: requests
  window: 60s
  description: Default limit for listing jobs or getting a job by ID.
- scope: per-team-per-endpoint
  plan: Enterprise
  limit: 500
  unit: requests
  window: 60s
  description: >-
    Enterprise plans advertise "API rate limits (500 req/min)" on the pricing page, with custom limits
    and higher concurrency negotiable.
  source: https://www.pyannote.ai/pricing
response_headers:
  on_success:
  - header: X-RateLimit-Limit
    meaning: Maximum requests allowed in the current window.
  - header: X-RateLimit-Remaining
    meaning: Requests remaining in the current window.
  - header: X-RateLimit-Reset
    meaning: Seconds until the current window resets.
  on_exhaustion:
  - header: Retry-After
    meaning: Seconds to wait before retrying.
  note: >-
    Uses the legacy X-RateLimit-* family rather than the IETF draft RateLimit-* headers. All three are
    documented as present on successful responses, so an agent can pace itself without first tripping
    a 429.
exhaustion:
  status: 429
  body: '{"requestId": "<uuid>", "message": "<error message>"}'
  spec_description: Too many requests
  spec_source: openapi/pyannoteai-api-openapi.yml
other_throttles:
- kind: streaming backpressure
  description: >-
    The WebSocket gateway enforces a maximum 5-second audio buffer; sending audio faster than real time
    closes the connection.
  source: asyncapi/pyannoteai-streaming-asyncapi.yml
- kind: budget cap
  description: >-
    A team that exceeds its configured monthly budget has subsequent API requests rejected. Enforcement
    may lag.
  source: https://docs.pyannote.ai/administration/billing
- kind: webhook delivery retries
  description: >-
    Outbound webhooks retry up to 3 times (immediate, +1 minute, +5 minutes), with x-retry-num and
    x-retry-reason headers on each attempt.
  source: https://docs.pyannote.ai/webhooks/receiving-webhooks