Google has introduced a major dual update to its consumer AI ecosystem, bringing the newly released Gemini 3.7 Flash model directly into the Gemini app while giving users direct control over visible branding on AI-generated creations. The update pairs substantial performance upgrades in code generation, agentic reasoning, and interface design with a practical usability feature: the ability to turn off the signature sparkle corner watermark on generated images, videos, and audio tracks.
This release marks an important shift in Google’s product strategy, balancing creative freedom for professionals with invisible provenance tracking to preserve digital content integrity.
Gemini 3.7 Flash Arrives in the Gemini App
Positioned as the high-efficiency agentic workhorse in Google’s Gemini 3 model family, Gemini 3.7 Flash is designed to deliver pro-tier reasoning and execution speeds at a fraction of typical computational overhead. Now available for daily conversations and complex tasks inside the Gemini mobile and web apps, the model introduces noticeable improvements over its predecessors in several critical domains:
- Advanced Software Engineering: The model demonstrates improved performance on real-world coding benchmarks, resolving software bugs, handling terminal commands, and producing reliable scripts on the first attempt.
- Web Development & UI Generation: Gemini 3.7 Flash interprets UI design references more faithfully, building complete interactive layouts, front-end components, and web applications with fewer iterative prompts.
- Document Comprehension & Knowledge Work: Processing dense PDFs, tabular spreadsheets, and multi-modal materials is faster and more contextually accurate, making the model effective for research and data extraction.
For everyday consumers, this upgrade translates to snappier, more precise answers. For developers, creators, and analysts, it establishes the Gemini app as a capable workbench for drafting production-grade code and synthesising large volumes of unstructured data.
Powering Gemini Spark and Agentic Productivity
Beyond standard chat prompts, Gemini 3.7 Flash serves as the underlying engine for Gemini Spark, Google’s personal AI assistant integrated into Google Workspace.
Agentic systems differ from traditional chatbots because they can plan multi-step workflows, autonomously call necessary software tools, and execute sequential tasks with minimal manual intervention. Powered by Gemini 3.7 Flash, Spark can:
- Coordinate Across Google Workspace: Seamlessly interact across Gmail, Google Docs, Drive, and Sheets to gather context, draft structured deliverables, and perform multi-file summaries.
- Execute Multi-Step Instructions: Follow complex workflows with fewer retries, reducing the need for constant user supervision.
- Automate Everyday Workflows: Handle repetitive administrative and back-office tasks, helping professionals save meaningful time during daily knowledge work.
New Media Watermark Setting: Removing the Visible Sparkle
Alongside the intelligence upgrade, Google has addressed a persistent request from designers and marketers by introducing a dedicated Media watermark toggle.
Historically, images generated with Imagen models, video renders from Veo, and musical tracks made with Lyria have featured a small, visible sparkle glyph in the lower-right corner. Under the new setting, users can turn off this visible graphic overlay across all newly generated visual media.
Note: The visible watermark toggle remains switched on by default. Users who wish to export clean assets must manually disable it within their account preferences.
Step-by-Step: How to Turn Off the Visible Watermark
Disabling the visible overlay takes only a few seconds on both mobile devices and desktop browsers:
- Open the Gemini mobile app on your Android or iOS device, or navigate to gemini.google.com.
- Tap the Menu icon (three horizontal lines) in the top-left corner and select Settings.
- Locate the Media watermark option in the settings menu.
- Toggle the switch to Off.
Once turned off, all subsequent visual media generated in the session will be produced without the corner sparkle icon.
Invisible AI Identification: SynthID and C2PA Remain Intact
A crucial distinction in this update is that removing the visible watermark does not make generated media untraceable. Google maintains a dual-layer provenance architecture to ensure digital transparency and prevent malicious deepfakes or misattribution:
- DeepMind SynthID: All images, video clips, and audio tracks generated by Gemini automatically embed an imperceptible, tamper-resistant digital watermark within the raw data pixels and audio waves. SynthID survives compression, cropping, screen grabs, and colour filtering.
- C2PA Content Credentials: Gemini exports cryptographic metadata conforming to the Coalition for Content Provenance and Authenticity (C2PA) open standard. This allows third-party platforms, search engines, and social networks to verify the asset’s AI origin independently.
While the outer canvas remains clean for professional use, the underlying cryptographic signals remain firmly attached to every file.
Availability, Regional Rules, and Account Constraints
The rollout of the Media watermark setting is subject to specific geographical and account-level guidelines:
- Enterprise & Education Accounts: If your Gemini access is managed through a Workspace business, enterprise, or school account, the toggle is currently unavailable, and visible watermarks remain enforced by policy.
- Regional Regulatory Frameworks: In jurisdictions with strict transparency mandates regarding AI markings, access to the toggle may be restricted to Google AI Ultra subscribers or kept active to comply with local laws.
- Phased Rollout: The setting is deploying progressively across the Gemini app, web client, and Google Flow video tools.
What This Means for Everyday Users and Creators
The combination of a smarter model and customizable export options delivers distinct advantages across multiple fields:
- Content Creators & Designers: Export clean, presentation-ready graphics and concept art without needing manual post-crop editing or cleanup in image editors.
- Software Developers: Rapidly prototype front-end web interfaces and debug multi-file scripts using faster, more accurate reasoning capabilities.
- Marketers & Small Businesses: Produce clean promotional collateral, storyboards, and audio snippets directly suited for publication and client presentations.
- Students & Researchers: Summarise complex technical papers and extract structured insights from large document collections with greater precision.
Frequently Asked Questions (FAQs)
Does disabling the visible watermark remove SynthID protection?
No. Turning off the visible sparkle toggle only removes the graphic overlay from the surface of the image or video. DeepMind’s invisible SynthID watermark and C2PA metadata remain embedded within the file data.
Is Gemini 3.7 Flash free in the Gemini app?
Yes, core chat and task features powered by the Gemini 3.7 Flash model are available to general users on the web and mobile apps, while advanced Workspace agent capabilities via Gemini Spark are included in Google AI Pro and Ultra subscription plans.
Why is the watermark toggle missing from my Gemini settings?
If you cannot see the setting, your account may be managed by an employer or school, or you may be located in a region where visible AI disclaimers are legally mandated. It may also still be propagating to your device during the phased global rollout.
Does the setting work on existing, previously generated media?
No. The toggle applies only to new media generated after the setting is switched off. Past generations already saved with a visible watermark will retain their original appearance.
Also Read: OpenAI Expands GPT-5.6 Luna Access: What ChatGPT Free Users Need to Know
Conclusion
With the integration of Gemini 3.7 Flash and an optional visible watermark setting, Google is refining the Gemini app into a more versatile tool for daily work and creative output. By delivering faster coding, reliable agentic automation, and clean visual media outputs while maintaining rigorous invisible provenance, the platform provides creators and developers with professional-grade capabilities without sacrificing ethical AI safeguards.