How to request Zero Data Retention (ZDR) for a Gemini Developer API project?

Hello,

We use the Gemini Developer API with a billed API key (paid tier) from Google AI Studio, for a small business app that processes sensitive documents. We would like to enable Zero Data Retention (ZDR) for our project, specifically opting out of the abuse-monitoring prompt logging, so that prompts/responses and identifiable metadata are cleared prior to logging, as described in the official docs: https://ai.google.dev/gemini-api/docs/zdr

Our integration already meets the ZDR requirements: no explicit context caching (cached_content), no File API persistence, no Grounding with Google Search/Maps, and no Interactions API state storage.

Questions:

  1. What is the exact process to request ZDR approval for a Gemini Developer API project? Is there a request form, or is invoiced billing required?
    1. Is ZDR available for small / non-enterprise paid-tier accounts on the Developer API?
  2. I can provide the project ID privately if needed. Thank you!

Following up — is there any official guidance from the Google team on the ZDR request process for Developer API projects? Still hoping for a reply. Thank you!

Quite critical for us too

Hi @T_tax and @Theo_Matussiere

I have DM’ed you for more details

Hi Mustan,

Thank you for getting back to me, and thank you for the DM.

Yes, we are currently considering migrating to Vertex AI, as it was also recommended to us by the Google Cloud sales team. We understand that Vertex AI does not use customer data for training and offers stronger data governance, which fits our requirements as a professional firm handling confidential client documents.

Before we proceed, I would appreciate your guidance on two points:

  1. On Vertex AI, is there an official process to request an exemption from abuse-monitoring data logging (i.e., zero data retention)? If so, could you point me to the application route or documentation?

  2. Is there anything else we should be aware of when migrating from the Gemini Developer API to Vertex AI (e.g., differences in API endpoints or model availability)?

Thank you again for your support.

Best regards,
T_tax