generated: '2026-08-17'
method: searched
source: https://docs.pyannote.ai/ + openapi/pyannoteai-api-openapi.yml
versioning:
  scheme: uri-path
  current: v1
  docs: https://docs.pyannote.ai/api-reference/test
  note: >-
    Endpoints are served under /v1/ with one exception: the team job list is GET /v2/jobs
    (operationId getJobsByTeamV2), versioned forward on its own. No published policy explains the split
    or the fate of any /v1 jobs-list predecessor.
  spec_version_field: >-
    info.version in the published OpenAPI is the literal string "local", not a release identifier — the
    spec carries no usable version number for change tracking.
  asyncapi_version: 0.0.1-beta
  asyncapi_note: >-
    The streaming WebSocket gateway is explicitly labelled beta in its own AsyncAPI info.version.
deprecation:
  policy_url: null
  policy_published: false
  sunset_header: unknown
  note: >-
    No deprecation policy, no sunset/deprecation notice page, and no RFC 8594 Sunset/Deprecation header
    documented anywhere in the docs or the specs. No Deprecation pointer is emitted in apis.yml because
    the provider publishes no such policy.
deprecated_operations: []
deprecated_operations_note: >-
  Zero operations in the OpenAPI carry deprecated: true.
changelog:
  published: false
  probed:
  - url: https://docs.pyannote.ai/changelog
    status: 404
  - url: https://docs.pyannote.ai/changelog.md
    status: 404
  note: >-
    No dated API changelog. The llms.txt documentation index lists no changelog/release-notes page. The
    only dated change record found anywhere is the release history of the Python SDK on PyPI
    (7 releases, latest 0.4.0 on 2026-01-14). No ChangeLog pointer is emitted.
status_page:
  url: https://status.pyannote.ai
  http_status: 200
  docs: https://docs.pyannote.ai/support/status
  note: >-
    Publicly reachable status page covering current service status, ongoing issues and scheduled
    maintenance. Linked from the docs support section.
sla:
  url: null
  uptime_target: null
  note: >-
    No published uptime SLA. Enterprise plans advertise "custom limits, higher concurrency" and
    dedicated Slack support, but no availability commitment is stated publicly.
support:
  faqs: https://docs.pyannote.ai/support/faqs
  troubleshooting: https://docs.pyannote.ai/support/troubleshooting
  email: support@pyannote.ai
  enterprise: Dedicated Slack channel (Enterprise plan)
data_lifecycle:
  docs: https://docs.pyannote.ai/data-retention
  uploaded_media_retention: 48 hours
  job_output_retention: 24 hours
  stream_audio_retention: none (never stored; processed on ephemeral workers)
  stream_metadata_retention: retained for billing and debugging (model used, duration in seconds)
  model_training: >-
    "We never use your audio data, outputs produced by jobs or streams, or any other customer data to
    train our AI models." Example data may be used for validation/testing only with explicit consent.
  note: >-
    Data retention is unusually well documented and is a genuine operational-lifecycle strength. It is
    also an integration constraint an agent must respect: results vanish 24 hours after completion, so
    outputs must be persisted client-side.