The short version: we never store what you write or what the model writes back. Here's exactly what we do keep, and why.
The content of your messages and the model's replies. They pass through our systems to generate a response and are never written to a database, never logged, never used to train anything, and never reviewed by a person as a matter of course.
| Data | Why |
|---|---|
| Your email address | To log you in and send account/verification emails |
| Your API key | So the gateway can check who's calling it |
| Token counts and timestamps per request | To calculate and deduct your credit balance |
| Payment records (amount, status, date) | To confirm your credit purchases went through |
Generating a reply requires briefly sending your message to a GPU server we rent from a third-party infrastructure provider (RunPod). We configure that connection to delete each request as fast as their platform allows once it's answered, rather than the default retention period they'd otherwise apply. We don't control their systems directly, so this is our best practical guarantee rather than an absolute one.
Every message is automatically screened before it reaches the model. Content indicating sexual content involving minors is blocked outright and never processed. This check happens regardless of any other privacy setting.
You can delete your account and stop using the service at any time. Because we don't keep chat history, there's nothing to "export" — the only data tied to you is your account details and the usage numbers described above.