Privacy
Privacy Notice
Last updated: July 15, 2026
This notice explains how Zro handles data for its sovereign open-source model inference service.
What we process
We process prompts, completions, web-search queries, API metadata, token counts, billing events, and account details as needed to provide inference and tools, operate accounts, measure usage, collect payment, support reliability, and provide any logging or history features you choose to enable.
Prompt and request retention
By default, we do not retain prompt text, completion text, or request bodies after the inference request is processed. If you enable features such as chat history, request logs, debugging logs, or exports, we may store the related content and metadata for as long as needed to provide those features or until you delete or disable them, subject to any agreed retention settings.
Training use
We do not use customer prompts, completions, request bodies, chat history, or logs for model training, fine-tuning, evaluation datasets, or product analytics unless you separately and explicitly agree to that use.
Infrastructure location
We operate inference on EU infrastructure. Account, billing, authentication, and operational vendors may process limited metadata required to provide their services.
Web search
When an agent invokes Zro web search, Zro sends the search query and the parameters needed to perform that search to the Brave Search API. We do not send the full conversation to Brave Search. Brave may process the query and related request metadata to return search results.
Billing and account records
We retain account identifiers, API key metadata, billing status, token totals, and usage-event records needed for metered billing, fraud prevention, accounting, support, and service reliability.
Contact
For privacy questions or deletion requests, contact the Zro team through the support channel associated with your account.