Runware Zero Data Retention Policy for LLMs

Last updated: September 15, 2026

Zero Data Retention (ZDR) is an organisation-level option for Runware’s text inference (LLM) API.

ZDR applies only from the moment it is enabled on the Runware organisation UUID and does not affect any data processed prior.

This document describes what we do and do not retain when it is enabled.

Data	How we handle it
Prompt and generation data	Not retained. Processed transiently in volatile memory for the duration of the request, and not kept beyond the time required to generate and deliver the output. If prompt caching is active, some prompt data (and associated KV caches) can be stored in volatile memory. This cache is never written to disk, evicted automatically, and discarded on replica restart.
Technical metadata	Retained for billing, debugging, reliability monitoring and security auditing.
Billing records	Retained as required by applicable tax and accounting law.
Account and authentication data	Retained for the duration of the account plus any statutory retention period, for service delivery and legal compliance.

We do not use any data submitted through our API to train or improve models.

Scope

Zero Data Retention applies to the text inference models listed below.