This page outlines the storage and management of Gemini API logs, which are developer-owned API data from supported Gemini API calls for projects with billing enabled. Logs encompass the entire process from a user's request to the model's response.
1. Data that can be shared
As a project owner you have the choice to opt-in to logging of Gemini API calls, for your own use or for feedback and sharing with Google to help us continually improve our models.
With logging enabled, you can help us build AI systems that continue to be valuable for developers across various fields and use cases by choosing to contribute the following data for product improvements and model training:
- Datasets: Use the Logs and Datasets interface of Google AI Studio to choose logs (requests, responses, metadata etc.) of interest from supported Gemini API calls; contributed through inclusion in datasets, with the option to opt-out during dataset creation.
- Feedback: When reviewing logs, you can provide feedback; including thumbs up/down ratings and any written comments you provide.
When you share a dataset with Google, your logs in that dataset, including requests and responses, will be processed in accordance with our Terms for "Unpaid Services," meaning the dataset may be used to develop and improve Google products, services, and machine learning technologies, including improving and training our models. Do not include personal, sensitive, or confidential information.
2. How we use your data
Logs will expire after 55 days by default. They will become unavailable after this period. Datasets can be created to retain logs of interest or value beyond this period for downstream use cases and optional contribution to model improvements. Logs stored in datasets do not have set expiry dates, however each project has a default storage limit of up to 1,000 logs.
By default, because logging is only available for billing-enabled projects, prompts and responses within logs are not used for product improvement or development, in accordance with our Terms on data use.
If you choose to share datasets of your logs with Google, those datasets will be used as real-world demonstration data to better understand the diversity of domains and contexts AI systems and applications are used in. This data may be used to improve model quality, and inform the training and evaluation of future models and services. This data is processed in accordance with our data use terms for Unpaid Services. Accordingly, human reviewers may read, annotate, and process the API inputs and outputs you share. Before data is used for model improvement, Google takes steps to protect user privacy as part of this process. This includes disconnecting this data from your Google Account, API key, and Cloud project before reviewers see or annotate it.
3. Data permissions
By opting-in to contributing API data, you confirm that you have the necessary permissions for Google to process and use the data as described in this documentation. Please do not contribute logs containing sensitive, confidential, or proprietary information obtained through the paid service. The license you grant to Google under the "Submission of Content" section in the API Terms also extends, to the extent required under applicable law for our use, to any content (e.g., prompts, including associated system instructions, cached content, and files such as images, videos, or documents) you submit to the Services and to any generated responses.
4. Data sharing and feedback
You can help us advance the frontier of AI research, the Gemini API and Google AI Studio by opting in to share your data as examples, enabling us to continually improve our models across various contexts and build AI systems that continue to be valuable to developers across various fields and use cases.