Skip to content
checkaifree
AI image detector

Gemini and Nano Banana image checker

Check if an image is AI-generated when it may be from the Gemini app or a Nano Banana model. Nano Banana is Google’s name for the current Gemini image models, not a separate mystery generator.

Checking scope
Files
JPEG, PNG, WebP · up to 10 MB
Model check
General AI-image estimate, when available
On this page
Gemini app and Nano Banana files
Signatures
Not verified by this site
Specimen plateAwaiting image

Drop an image here

or choose a file

Metadata demos (controlled fields):
JPEG · PNG · WebPUp to 10 MBNo sign-up
How the file travels

What Google attaches, and what survives

Google’s Gemini image documentation says all generated images include a SynthID watermark. SynthID is invisible and lives in the pixels. Google’s SynthID detector, not this site, is the tool that reads it. A screenshot can still be a Gemini image. We will not see the watermark either way.

Some current Gemini image models document C2PA next to SynthID. Google Cloud’s Nano Banana 2 note describes SynthID together with Content Credentials so a file can say how AI was used, not only that it was used. That metadata is in the original file. A messenger recompress or a screenshot drops it. Imagen is the previous Google image family: the Gemini API docs say Imagen is shut down there and point developers to Nano Banana. Old Imagen files belong on the Imagen 3 guide.

The Gemini app, AI Studio, and the API can hand you different containers. Keep the original download. A photo of a screen, or an image a chat partner forwarded, is usually only pixels. Our optional model check is a general AI-image estimate. It is not a Gemini or Nano Banana classifier, and it is not a SynthID result.

The drawings are original illustrations of file paths. They are not outputs from Gemini.

Diagram of a Gemini original with SynthID and C2PA versus a forward that keeps only pixelsGemini fileForwarded copyMetadata dropped
Google’s watermark is in the pixels. Content Credentials, when that model wrote them, are in the file and are easy to lose.
Gemini evidence layers: SynthID, C2PA, general model scoreSynthID — always on, not read hereC2PA — some current modelsOur score — general only
We can only offer the general score. SynthID and signed credentials are Google’s tools.
Gemini and Nano Banana page versus the legacy Imagen pageHereOther guideNano BananaGemini appOld ImagenSynthID decode
Imagen is the legacy name. Current Gemini app images belong here.
Metadata and credentials

Metadata and Content Credentials

Two official signals, with different durability. SynthID is the one Google says is always on for these generated images. C2PA is documented for some current Nano Banana models and enterprise outputs, and it is the fragile one. Neither signal is something this checker validates. A string match on “c2pa” or a Google tool name is not a trusted manifest.

Google also says SynthID is not a judgment about whether the picture’s story is true. A watermarked illustration of a real event is still an illustration. A file with no watermark and no manifest is not thereby a photograph.

This site can notice source text in the first part of a file. It does not verify C2PA signatures, asset bindings or certificate trust, and it has no SynthID decoder.

Look twice, then don’t conclude

Visual tells, treated cautiously

Nano Banana is marketed for text, editing and consistency. Those are product claims, not visual fingerprints.

Text in the picture

Current Gemini image models are built to render labels and diagrams. So are other 2026 models, and so are human layout tools.

Edited photos

People use Gemini to change a real photo. The result can be a mix. A camera EXIF block may be the original’s, the edit’s, or neither. Do not treat leftover EXIF as proof the whole frame is untouched.

No visible watermark to hunt

SynthID is not a logo in the corner. Looking for a mark and not finding one tells you nothing about Gemini.

How this check handles the file

Source clues are checked in your browser first. When needed, file-header data and a smaller image copy may be sent to our backend and detection provider. Model copies have embedded metadata removed. Image URLs are retrieved by our server; see the privacy policy for processing and provider retention. Privacy details

Sources for the generator notes

Frequently Asked Questions

Questions about Gemini.

The answers below are about this generator’s files, not a promise to name the model.

What is Nano Banana?

It is Google’s product name for Gemini image models, including later Flash and Pro image variants. This page covers that family. It does not identify which Gemini snapshot produced a file.

Can you read the SynthID watermark?

No. This site has no SynthID decoder. Use Google’s own detector if you need that signal. Our optional score is a separate, general model estimate and can be wrong or unavailable.

Does Gemini write C2PA?

Google documents SynthID on generated images, and documents C2PA as well on some current Nano Banana models. A given screenshot may contain neither that we can see. We do not verify signatures.

Is the check free?

No account is required. Availability and request limits apply.

Related reading

These notes explain what a file can and cannot show. They do not name Gemini from the pixels.

Explore Other Generators

Also inspect other generative models

All checking guides
Image review scenarios

Reviewing profile photos or art entries?

Review LinkedIn candidate photos or competition artwork with context, and read our method and limitations.

Method and limits