Free Tier data handling on model retries

Gemini API
1

Hello,

We used the Gemini Developer API Free Tier with retry logic across multiple Gemini models. As a result, the same input data may have been sent to multiple models.

Could you please clarify:

  1. Is input data stored separately for each model or each request?
  2. Is the input data used for training or improving each individual model, or for Gemini services overall?
  3. If the same data is retried across multiple models, does this increase the scope of its use for training or model improvement?
  4. How long is input data retained when using the Free Tier?

This is for an internal security review.

Thank you.