Gemini API Logging Is Now On by Default for New Paid Projects: What Agencies Should Check

Gemini API logging is now on by default for new paid projects. Google’s Gemini API changelog says that since October 26, 2026, new paid-tier projects store Generate Content requests by default. Existing projects keep their current setting.

This matters most for agencies that create a new Google AI Studio project for each client. Any project created on or after October 26 now keeps prompts and responses in its logs unless someone changes the setting. This piece covers what is logged, how long it is kept, who can see it, how to turn it off, and the Gemini features that keep data no matter what the logging setting says.

The short version: new paid projects now log requests, responses and metadata for up to 55 days by default. Google says these logs are not used to improve its products unless you share them as a dataset. You can turn logging off for the whole project in AI Studio, or per request with store=false. Some features, such as Grounding with Google Search, keep data for 30 days with no way to turn it off. Check every project created since October 26, especially those that handle donor or member data.

What changed on October 26

The changelog entry covers three Generate Content methods: generateContent, streamGenerateContent and batchGenerateContent. For new paid-tier projects, these now store requests by default with store=true. Google says this matches how the Interactions API already worked.

Two points from the entry are easy to miss:

• Existing projects are unchanged. A project created before October 26 keeps whatever setting it had.
• The change only affects paid-tier projects. Google’s logging guide says log storage is only available for projects on the paid tier.

Google’s Logs and datasets guide, updated October 26, adds one detail. Older projects keep their existing setting, which is off unless someone turned it on in AI Studio.

So a client project from last summer probably does not log. A project you created last week probably has Gemini API logging on.

What Gemini API logging stores and for how long

Google’s logging policy page says the logs cover the full path from a request to the model’s response. That includes the request, the response and metadata.

The policy describes these logs as owned by the developer and private to your Google Cloud project. They are separate from the logs Google keeps for abuse monitoring.

On retention, the policy states: "Logs are retained for a default maximum period of 55 days." You can shorten that window in AI Studio to 7, 14 or 28 days. Logs you save into a dataset do not have a set retention period.

The logging guide lists which calls get logged:

• generateContent, streamGenerateContent and batchGenerateContent
• Interactions API calls, except Managed Agents
• Calls made through the OpenAI compatibility endpoints

That last item matters for agencies. If a client tool talks to Gemini through OpenAI-style code, its calls can still land in Gemini API logging.

Who can see the logs and whether Google uses them

The policy says logging is tied to billing-enabled projects. Because of that, Google says prompts and responses in these logs are not used for product improvement or development by default.

That changes if you share. AI Studio lets you build datasets from selected logs and contribute them to Google. Per the policy, shared data may be used to improve model quality and to train and evaluate future models. Human reviewers may read shared inputs and outputs. Google says it disconnects the data from your Google Account, API key and Cloud project first.

The policy also describes a Gemini API Contributor Tier, an invite-only private preview. It offers up to 50% off token costs on some models in exchange for permission to use that data. For client work with personal data, read those terms closely before joining.

Inside your own team, think about access. The logs live in the Cloud project. Check who has access to each client project before you leave logging on.

How to turn Gemini API logging off

There are two ways, per Google’s pages.

For a whole project. The policy says: "As a project owner, you can turn logging on or off at any time from the Logs and Datasets page in AI Studio." The logging guide says the Settings panel has separate toggles for Generate Content and the Interactions API.

For a single request. The changelog says you can set store=false on individual requests. The logging guide adds that a request-level store value overrides the project setting. It also notes that sending store on Generate Content requests needs a recent SDK version, so check your SDK before relying on it.

If you keep logging on, shorten the window. Seven days is enough for most debugging. You can also export logs as CSV or JSONL files or to Google Sheets, so plan where those exports go.

What Gemini keeps even with logging off

Google’s zero data retention page, last updated September 14, 2026, is the place to check this. It explains that zero data retention on the Gemini Developer API is not one switch. Some features keep data on their own schedule.

For paid services, the same page says Google does not use your prompts or responses to improve its products. It also points to Vertex AI for guaranteed zero data retention. Our guide on moving from AI Studio to Vertex AI covers that path.

How OpenAI and Anthropic handle API logs

The other two large API platforms also keep some data by default. OpenAI’s data controls guide says API data is not used for training unless you opt in, and abuse monitoring logs are kept for up to 30 days. OpenAI’s API reference says the Responses API store setting defaults to true and keeps response data for at least 30 days. Anthropic’s privacy center says it deletes API inputs and outputs within 30 days, with exceptions such as the Files API and usage policy enforcement. Both offer zero data retention only by agreement or approval. For key hygiene on the OpenAI side, see our post on OpenAI API key expiration.

A Gemini API logging checklist for client projects

Use this list for each client project in AI Studio:

• Find new projects. List every paid project created on or after October 26, 2026. Those log by default.
• Decide per project. Turn logging off where prompts may include donor, member, student or staff data.
• Shorten retention. If logging stays on, set the window to 7 days unless you need more.
• Set store=false in code. Do this for sensitive requests, so the choice survives a project setting change.

Then look at the features around the model call:

• Check grounding. If a tool uses Grounding with Google Search, note the 30-day storage in your client documentation.
• Review access. Check who can open each Cloud project and its logs.
• Do not share datasets by accident. Sharing a dataset with Google allows training use and human review.
• Write it down. Record the setting, retention window and date for each client.

If you are new to AI Studio projects and keys, start with our Google AI Studio 101 guide. Teams that started a project for Gemini 4 Argon after October 26 should check that project first. For the Workspace side of Gemini records, see our post on Google Vault Gemini exports.

Frequently Asked Questions