Gemini Ditches Visible Watermarks, Boosts Performance with 3.7 Flash Upgrade

Google Gemini has rolled out a new feature that allows users to disable visible AI watermarks on generated images, videos, and audio clips. This update comes after Google first teased the setting earlier this summer.

The toggle is available in the settings gear icon at the bottom-left corner of the web app, under the 'Media watermark' section. Turning off the toggle removes the familiar corner spark icon but doesn't affect the underlying security layers.

Google is keeping invisible SynthID watermarks and C2PA provenance metadata embedded directly into the file, ensuring automated systems can verify the content's AI origin. This dual approach provides creators with clean visuals for presentation decks or creative projects while maintaining security.

The update also includes a major backend upgrade to Gemini 3.7 Flash, which replaces version 3.6 Flash across Android, iOS, web, and desktop apps. The new model powers the Gemini Spark agent for subscribers on AI Pro and Ultra plans, bringing performance jumps in coding, UI design generation, and complex document analysis.

The 3.7 Flash model scored 65.3% on the DeepSWE v1.1 bug-fixing test, a significant improvement from the previous version's 49.0%. Additionally, Google has cut introductory API pricing in half through the end of the year, making 3.7 Flash a budget-friendly option for developers building multi-step automated workflows.