Today, we’re introducing major enhancements to how you can bring your own data into the Gemini API, making it easier and faster to integrate diverse datasets into your AI applications. New support for Google Cloud Storage (GCS) object registration, external HTTPS and signed URLs, and significantly increased inline file size limits are now available—removing previous bottlenecks and enabling seamless data ingestion from existing storage systems. Previously, working with large files such as videos, long audio recordings, or extensive documents required uploading them to the Gemini Files API, where they were only retained for 48 hours. While suitable for initial prototyping, this temporary storage model proved impractical for production applications that depend on persistent, reliable access to data stored in cloud environments. To address this, we’ve introduced two new input methods that eliminate the need to re-upload or move data: External URLs (public and signed): The Gemini API now supports direct access to files hosted on public domains or private storage via signed URLs. During processing, the API securely retrieves the content on-demand, so you no longer need to download files to your backend before sending them to the API. This improves efficiency, reduces latency, and simplifies infrastructure. Register GCS files directly: If your data already resides in Google Cloud Storage, you can now register those files directly with the Gemini Files API without transferring or copying them. This preserves data integrity, reduces bandwidth usage, and accelerates deployment for applications built on Google Cloud. In addition, we’ve expanded the maximum size for inline file uploads—ideal for quick prototyping, real-time interactions, and handling larger media. The new limit increases from 20MB to 100MB for base64-encoded data, with minor variations depending on the data type. This allows developers to work with higher-resolution images, longer audio clips, and more complex input formats without relying on external storage. Together, these updates give you a flexible, scalable, and production-ready toolkit for data ingestion. Whether you're using public links, private URLs, or data already in GCS, or prefer the simplicity of inline uploads, the Gemini API now supports your workflow at every stage—enabling faster development, smoother deployment, and more powerful AI applications.