changelog

Google Rolls Out Gemini 3.7 Flash to Chat Interface, Adds Watermark Toggle

TL;DR

Gemini 3.7 Flash has replaced 3.6 Flash in the Gemini app's model picker on Android, iOS, web, and macOS, following its debut in the Spark agent a day earlier. The app also now lets users turn off visible corner watermarks on AI-generated images, video, and music.

3 min read
0

Google has expanded availability of Gemini 3.7 Flash from the Spark agent into general chat, while separately rolling out a long-requested option to disable visible watermarks on generated media.

3.7 Flash reaches the model picker

Gemini 3.7 Flash launched Wednesday morning powering only the Spark personal agent inside the Gemini app. Late Thursday, Google pushed the model into the main chat interface, where it replaced 3.6 Flash in the model picker across Android, iOS, web, and macOS.

The model picker now includes 3.7 Flash, 3.5 Flash-Lite, and 3.1 Pro. Notably, 3.5 Pro did not receive an availability update in this rollout.

According to Google, 3.7 Flash delivers improved reasoning and accuracy in "knowledge-dense fields like finance, law, and biosciences," along with gains in coding, web development, and UI generation. Google has not published specific benchmark scores for these claims. Within Gemini Spark, the company says 3.7 Flash improves tool use for Google Workspace apps, making the agent more efficient at completing tasks.

No pricing changes were disclosed alongside this rollout, and context window specifications for 3.7 Flash were not included in Google's announcement.

Watermark toggle arrives after July preview

Separately, the Gemini app now includes a setting to disable the visible corner watermark that appears on AI-generated media. The spark icon watermark shows up on images from Nano Banana, videos from Omni, and songs from Lyria. Google first previewed this control in July.

On the web version of Gemini, users can find the option by clicking the settings gear icon in the bottom-left corner, which now surfaces a "Media watermark" section with "On" and "Off" toggles. The visible watermark remains on by default.

Importantly, turning off the visible marker does not remove all provenance tracking. Invisible SynthID watermarks and C2PA metadata stay embedded in the file regardless of the visible-watermark setting. Google is also withholding the off-switch in "countries where it's required by law to keep" visible watermarks, though the company did not specify which jurisdictions.

The watermark control has already reached Flow, Google's video creation tool, and the company says Search will get the same option next. Both the 3.7 Flash rollout and the watermark toggle are rolling out gradually to all platforms over the coming days, meaning not all users will see the changes immediately.

What this means

The watermark toggle addresses a common complaint from users of AI-generated content who found the corner spark icon distracting for professional or creative work, though the persistence of invisible SynthID and C2PA metadata means provenance tracking remains intact for detection tools and platforms that check for it. The regional restriction signals Google is treating this as a compliance-sensitive feature, likely tied to EU AI Act transparency requirements or similar disclosure laws rather than a universal design choice. For 3.7 Flash, the staggered rollout — Spark agent first, then general chat a day later — reflects Google's now-standard pattern of testing new model versions in narrower, more controlled product surfaces before wider exposure. The exclusion of 3.5 Pro from this update batch suggests Google is managing model tiers on separate release cadences rather than pushing simultaneous version bumps across its entire Gemini lineup.

Comments

Loading...