The honest answer some searches deserve
Not every page here exists to convert a visitor into a tool user. If your image came from Google AI Studio or the Gemini API directly, the honest answer to "does this watermark my images" is: it applies SynthID, an invisible signal, but per current documentation it does not add the visible corner mark this tool removes. There may be nothing here for you to use.
What the tier behavior actually is
Per Google's own tier documentation, the split is not "does Google watermark AI images" but "which watermark, on which tier."
| Tier or access path | Visible mark this tool targets | What to expect |
|---|---|---|
| Free tier (Gemini app) | Supported | Visible corner mark plus SynthID. In scope for this tool. |
| Google AI Pro (Gemini app) | Supported | Visible corner mark plus SynthID. In scope for this tool. |
| Google AI Ultra (Gemini app) | Not supported | SynthID only, per current documentation. No visible mark for this tool to find. |
| Google AI Studio / Gemini API output | Not supported | SynthID only, per current documentation. Generated output ships without the consumer app's visible mark. |
That distinction comes directly from Google's Nano Banana Pro tier documentation, not from inference, and it is the single fact this page exists to surface clearly: Free and Google AI Pro get the visible mark and SynthID together; Google AI Ultra and direct AI Studio or API access get SynthID only.
Why that means this tool may not apply here
If an image only carries SynthID, there is no visible corner overlay for the detection engine to find, and it will correctly report nothing to remove. That is not a bug or a missed detection — it is the expected outcome for an access path that never applied a visible mark in the first place. See what this tool cannot remove and the SynthID comparison guide for the fuller picture of why SynthID sits outside every visible-mark tool's scope, this one included, regardless of which tier produced the image.
Which access path produced your image?
Answer the one that matches, to find out whether this tool has anything to do here.
Did you generate it in the free Gemini app or Google AI Pro? — Consumer tier
It likely carries both SynthID and the visible corner mark. The browser eraser is built for this case.
Did you generate it through Google AI Ultra? — Higher consumer tier
Per current documentation, Ultra output carries SynthID only. There is probably no visible mark here to remove.
Did you call the Gemini API or generate it in AI Studio? — Developer access
SynthID only, per current documentation. Building a visible-mark removal step into this pipeline is very likely solving a problem that was never actually present.
Not sure which path produced the image? — Check visually first
Zoom into the bottom-right corner at full size. If there is no semi-transparent logo there, this tool has nothing to do — SynthID is invisible either way.
Why the pricing question stays off this page
Google AI Ultra pricing shows up in several places with inconsistent figures across sources, and none of them are anchored firmly enough for this project to repeat with confidence. Rather than publish a number here that might already be wrong by the time it is read, this page anchors on the confirmed tier behavior — which tiers get SynthID only versus SynthID plus a visible mark — and leaves any price claim out entirely until it can be verified directly in a real, current billing flow.
Why a developer audience needs this page too
Developers building on the Gemini API sometimes assume their generated output must carry the same visible mark they have seen in the consumer Gemini app, and design an unnecessary post-processing step around that assumption. This page exists partly to correct that specific assumption directly, with a citation, rather than let it propagate through forum answers and half-remembered blog posts. If an integration only ever calls the API or AI Studio directly, the visible-mark handling this site's tool provides is very likely not something that integration needs, and adding it anyway is processing time spent on a problem that was never actually present in that pipeline.
If you do have a visible mark
Nothing above applies if your image actually shows a visible corner overlay — some workflows mix consumer-app exports with API calls, or pass images between tools, so a real mark can still show up on files people assume came from a watermark-free path. In that case, this project's normal coverage applies: check the size and position catalog for your exact image dimensions, and use the browser eraser the same way you would for any other qualifying Gemini output.
Last updated 2026-07-31