Model Rover

Data Privacy Policy

Understand retention windows and privacy commitments for request metadata and generated media.

The platform balances billing reconciliation and self-service troubleshooting with strict limits on how long request data is retained and how it is used.

Request metadata

The platform keeps request metadata needed for billing reconciliation, usage analytics, reliability, and support. Metadata can include the request ID, timestamp, model identifier, API key identifier, call mode, status code, timing metrics, usage meters, and final settled amount. Metadata does not contain prompt or output text.

Privacy and use commitments

For your request data, the platform commits to the following:

  • Never used for model training: We never use your prompts, request parameters, or model outputs to train, fine-tune, or improve any AI model, nor do we sell them or use them for commercial profiling.
  • Strictly controlled access: Your request logs are visible to your account for self-service debugging. Platform staff do not manually inspect your request content unless you submit a support ticket referencing the request ID for troubleshooting, or where strictly required for security abuse investigation or legal compliance, with all administrative access permission-gated and audit-logged.

Generated media

Image and video result URLs are temporary links and must not be used as permanent application storage:

  • Video generation results (retained for 48 hours): Video results and task lookup snapshots from API calls are temporarily retained for 48 hours, after which they are automatically and permanently deleted. Download and transfer the video to your own storage immediately upon receiving a completion callback or polling a succeeded status.
  • Image log preview: Image results are retained only briefly for preview in request-log details and are permanently deleted upon expiration.
  • Playground history: Test videos generated in the Playground are retained for 30 days before automatic physical deletion and must not be referenced by production applications.

Sensitive data and best practices

Because the gateway routes requests in real time to the upstream model provider you select, send only data you are authorized to process and redact sensitive fields on the client side where appropriate. Do not place plaintext passwords, API keys, private keys, payment credentials, or unnecessary sensitive personal data (PII) in prompts, request parameters, uploaded files, callback URLs, or metadata.

Help us improve this page

Found something unclear, outdated, or incorrect?

Last updated on