Skip to main content
Data retention defines how long Orq.ai keeps observability data before automatic deletion. Retention is tied to the subscription plan and cannot be changed from the UI.

​
Retention periods

The retention period applies per workspace. A legal duty, such as the six-month log retention required of deployers under the EU AI Act, sets a floor on the retention period. Request the period with the subscription. Enterprise plans retain traces and logs for six months or more, up to 365 days, which covers a six-month duty. Pay as you go retains 30 days, for duties with a shorter window.

​
What is covered

The following data types are deleted automatically after the retention period expires:
  • Traces and spans: trace records, span details, and associated metadata
  • Logs: OpenTelemetry log entries ingested through the platform
  • Insights: summaries, embeddings, insight annotations, and assignment records
  • Threads, messages, and responses: conversation data from the Responses API
  • Audit logs (Enterprise plans): organization activity records (see Audit Logs for details)

​
What is not covered

The following data types persist until manually deleted:
  • Datasets: dataset definitions and entries
  • Media and file attachments: files uploaded through the platform. Delete via DELETE /v2/files/{file_id}
  • Evaluator results: evaluation scores from the analytics engine
  • Annotations: annotation records on spans
When a trace expires, annotations referencing that trace become orphaned. The annotation queue shows a count of items whose trace no longer exists. Traces are removed when the retention period expires. See Erasure and Access Requests for the routes that erase workspace data and how to submit a data subject request.

​
Enterprise retention

Enterprise plans can request a specific retention period up to 365 days. Contact the account manager or support to adjust the retention window.