New Paid Gemini API Projects Now Log Prompts and Responses by Default. Google's Policy Keeps Those Logs Up to 55 Days Unless You Shorten It.
Google's Gemini API changelog for October 26 says new paid-tier projects now store Generate Content API requests by default, with store set to true for generateContent, streamGenerateContent, and batchGenerateContent. Existing projects keep their current setting. You can opt out per request or in the Logs and Datasets dashboard. Google's logs policy sets a default maximum retention of 55 days, with 7, 14, or 28 as the shorter options, and says ordinary logs are not used for training unless you contribute them as a dataset. We read the changelog and the policy page.
By T.W. Ghost
Google changed a default today, October 26, 2026, and put it in the Gemini API release notes rather than anywhere you would look. New projects on the paid tier now store Generate Content API requests by default: store is true for generateContent, streamGenerateContent, and batchGenerateContent, which brings those endpoints in line with the Interactions API. Existing projects keep whatever setting they had. If you created a paid Gemini project on or after today, assume your prompts and responses are being logged until you turn it off.
The retention side is on a different page. Google's data logging and sharing policy says logs are kept for a default maximum of 55 days, and that a project's window can be set to 7, 14, 28, or 55 days in AI Studio. Logs belong to the developer's project on billing-enabled accounts, and the policy says prompts and responses in ordinary logs are not used to improve models unless you contribute them as a dataset. We read the changelog and the policy page. Here is what held up.
What Changed
| Gemini API logging, October 26, 2026 | |
|---|---|
| Who | New projects on the paid tier |
| Default | store=true for generateContent, streamGenerateContent, batchGenerateContent |
| Existing projects | Keep their current setting |
| Opt out | store=false per request, or turn off Generate Content API logging in Logs and Datasets settings |
| Retention | Default maximum 55 days; project window selectable at 7, 14, 28, or 55 days (logs policy) |
| Datasets | Can keep selected logs beyond the window; contributing a dataset can send data to model improvement |
| Training | Ordinary logs not used for training unless shared via a dataset (logs policy) |
The flip is scoped, and the scope is the trap. Old projects are untouched, so a team that audited its settings in September is fine and a team that spun up a new paid project this week is logging. The two can sit in the same organization.
55 days is the ceiling, not a suggestion. The policy page lists it as the default maximum. Shorter windows exist; you have to pick one.
A dataset is a different door. Keeping a log past the window, or contributing it, is a deliberate act in the dashboard. The default change does not do that for you, per the policy as written.
Verified vs Unconfirmed: The Scorecard
| Claim | Verdict |
|---|---|
Oct 26: new paid-tier projects default to store=true on the three Generate Content endpoints | Verified (Gemini API changelog) |
| Existing projects keep their setting; opt out per request or in Logs and Datasets | Verified (changelog) |
| Default maximum retention 55 days; options 7, 14, 28, 55 | Verified (Gemini API logs policy) |
| Ordinary logs not used for training unless contributed as a dataset | Verified (logs policy wording) |
| Exact AI Studio control labels for retention | Unverified (policy describes the windows; confirm in the console) |
What To Do
Audit every paid Gemini project created on or after October 26: check store on your requests and the retention window in the dashboard, and set both deliberately.
Put the retention choice in your data map if you handle customer content; 55 days of prompt storage is a fact your privacy review needs to know, and 7 is available.
The frontier is elsewhere: nothing about the models moved. Gemini 3.8 Flash is still the Gemini model to measure, and our model comparison has the current standings side by side.
Sources
*Not sure which provider's defaults match your data rules? Take the free 2-minute quiz and get matched. Then read the other policy change this month that arrived without a model attached: OpenAI's usage policies for intermediaries.*