Zero Data Retention
Last updated 7 September 2026.
This page describes what the Auxilio API at api.auxilio.ai keeps about a request, and what it never keeps. It is part of our Privacy Policy.
What we do not keep
- Prompts, messages, system instructions, tool definitions, tool results and attachments you send.
- Completions, reasoning text and tool calls the model returns.
- Copies of any of the above in logs, caches, backups or training sets. Inference runs in GPU memory and the content is discarded when the response ends.
No third party receives request or response content. We run no analytics, tracing or logging service on API traffic, and we do not use customer content to train or evaluate models.
What we keep
Per request, for billing and abuse investigation:
- Timestamp, API key alias, model id, response status, latency.
- Token counts: prompt, cached prompt, completion and reasoning tokens.
- Client IP address and request path in the edge access log.
This metadata is kept for as long as it is needed to invoice usage and investigate abuse, and is deleted with the account. Account and billing records are covered by the Privacy Policy.
Where inference runs
On GPU hardware we own, in three datacenters in Romania, European Union. No inference is routed outside the EU and no external provider processes your content on our behalf. Marketplaces that resell the endpoint (for example OpenRouter) see your request under their own privacy terms before it reaches us.
Controller and contact
8000 Studio SRL, Romania, operates auxilio.ai and api.auxilio.ai. Questions and data requests: [email protected].