Gemini Avatar Feature Adds a 3D Digital Version of the User
Google is developing an Avatar feature for its Gemini AI assistant that would let users create a 3D digital representation of themselves and place it directly into AI-generated images and videos with a simple prompt tag. The feature surfaced in an APK teardown published by Android Authority on March 24, 2026.
The idea is straightforward: instead of uploading a photo every time, users would be able to call in their digital likeness directly while generating visual content. If released, the feature would expand Gemini beyond text-based personalization and into visual identity.
How the Gemini Avatar Feature Works
Face Scan Creates a 3D Model
The feature was discovered in version 17.11.54.sa.arm64 of the Google app by developer AssembleDebug. It builds on a "Characters" feature that was first spotted in February in an earlier beta.
Users would scan their face by recording a short video with their phone camera. That recording would then be used to generate a 3D model of the user.
Prompt Tags Like @me Could Insert the User Into AI Content
Once the model is created, it could be inserted into any Gemini image or video generation prompt by typing @me or @ followed by the user's username.
A string found in the app's code describes the feature this way: "Add yourself to any frame being created by typing @me or @%s in your prompt." The feature also appears to have a new "Avatar" option inside Gemini, along with a welcome screen that walks users through the video capture flow.
Gemini Avatar Feature Appears to Evolve From Earlier Characters Work
Earlier Beta Versions Pointed to the Same Direction
The Avatar feature appears to grow out of the earlier "Characters" menu found in version 17.4.66 of the Google app. That earlier discovery hinted at personalized digital characters, but at the time the recording function was not yet working.
The newer March finding suggests the feature has moved forward. The branding now appears simpler, using "Avatar," and the integration with @me points to a more direct use case for image and video generation.
Links to Android XR and Likeness Technology
Android XR's Likeness Feature Shows the Technical Foundation
The underlying technology appears connected to Android XR's "Likeness" feature, which launched in late 2025 for headset users creating digital avatars for video calls.
That system asked users to scan their face through a companion smartphone app. After that, the headset's sensors animated the avatar in real time to reflect facial expressions.
Gemini May Bring That Avatar Pipeline to Standard Android Devices
Google appears to be adapting that same pipeline for broader use through Gemini on regular Android devices. The shift would remove the need for an XR headset and bring the avatar creation process into a more mainstream setting.
Code tied to the earlier February discovery also referenced Android XR avatar functionality, reinforcing the connection between the two efforts.
Gemini Avatar Feature Is Still in Development
No Public Release Yet
The feature is not publicly available. And, like other findings from APK teardowns, there is no guarantee it will launch in its current form.
Google has not officially announced the feature and has not shared any release timeline.
What the Feature Could Mean for Gemini Personalization
If it does ship, the Avatar feature would mark a notable step in Gemini's personalization tools. Instead of limiting personalization to text-based context, it would let users bring their own visual identity into AI-generated images and videos through a reusable 3D model.
That would make it possible for users to appear in their own generated content without needing to upload a photo each time.