Pre-launchAPI access is by invitation. Prices shown are estimates.Request access →
Contact sales
ModelsPricingDedicatedDocsEnterpriseContact sales
Your data

Not stored. Not trained on. Not shared.What happens to your data

We run open models on infrastructure we operate. Your prompts and completions are processed in memory and discarded. This page is the plain-language version; the binding text is our Data Policy.

DataStored?Used for training?Retention
Prompts and inputsNo — memory onlyNoDiscarded when the response completes
Completions and outputsNo — memory onlyNoDiscarded after delivery
Request metadata (model, token counts, latency, status code, timestamp, API key ID)YesNo13 months, for billing and abuse prevention
Account data (email, name, billing details)YesNoWhile your account is active, then per our Privacy Policy

The only exceptions

  1. You ask us to debug a specific request and give us written permission to capture it. We delete it within 30 days after the issue is resolved.
  2. We're legally required to preserve specific records.
  3. Batch jobs, if you use them, are stored encrypted until results are delivered, then deleted.

Where we process data

Inference runs on GPU nodes we operate in Malaysia (Johor), Indonesia (Jakarta), and Ireland (Dublin). Requests are served from Johor by default, with Jakarta and Dublin as failover. See Regions and our sub-processors.

Questions

Email support@kurrens.ai or read our Privacy Policy.