Google has launched Gemini Avatar inside Flow AI, a feature that scans your face and voice to create a personalized AI clone you can insert into any AI-generated video. The system, which rolled out to paid subscribers throughout June 2026 following its announcement at Google I/O 2026 in May, solves one of AI video creation's toughest problems: keeping the same character looking consistent across multiple scenes without ever turning on a camera.

How Does Gemini Avatar Actually Work?

The technology combines three Google systems working together. Gemini Omni, Google's newest multimodal AI model announced at I/O 2026, powers the avatar creation. Veo 3.1, Google's video generation engine, renders the final video. And Flow AI, Google's creative workspace, stores your avatar as a reusable asset alongside other creative ingredients.

The process starts with a face scan using your phone's front camera. Unlike a flat photograph, the system captures your face's three-dimensional structure so your avatar looks genuinely like you in motion, not just in a still image. You then record your voice by reading a few sentences at your normal speaking pace and volume. Gemini Omni captures your voice's natural tone, rhythm, and characteristics. Once processed, which typically takes 3 to 10 minutes, your avatar is saved as a Flow AI asset and ready to use.

When you want to generate a video featuring your avatar, you simply type @me (or your saved avatar name) into any Flow AI video generation prompt. Veo 3.1 then creates the final clip with your avatar as the main character, complete with full motion and synchronized audio. Every video generated with your avatar includes an invisible SynthID watermark from Google DeepMind, marking the content as AI-generated for transparency and safety purposes.

What Are the Technical Requirements and Setup Steps?

Access to Gemini Avatar requires a Google AI Plus subscription at $19.99 per month, or higher tier plans. The feature is not available on the free tier. You'll need either the Google Flow AI mobile app or a desktop browser to access flow.google.

The setup process is straightforward but depends on environmental conditions. Good lighting during the face scan makes a significant difference; a well-lit room with natural daylight or a lamp directly in front of your face produces a substantially better avatar than a dark environment or backlighting. Similarly, a quiet environment during voice recording is essential, as background noise from air conditioning, traffic, or television affects voice clone quality.

Steps to Set Up Your Avatar in Google Flow AI

• Access Flow AI: Go to flow.google in your browser or open the Google Flow AI mobile app on your phone and sign in with your personal Google account.
• Locate the Avatar section: Look for the Avatar or Characters section in the project assets panel, the same area where saved Ingredients and Style references are stored. On mobile, it may appear as a dedicated button in the creation interface.
• Begin the face scan: Tap or click Create Avatar and position your face in the camera frame so it fills most of the frame. Follow on-screen instructions to look directly at the camera, slowly rotate your head left and right, and tilt your chin slightly up and down.
• Record your voice: Read the on-screen text at your normal speaking pace and volume. The recording typically takes 1 to 2 minutes of speaking, and you may be asked to read multiple passages to gather sufficient voice data.
• Wait for processing: Gemini Omni processes your data to build your avatar, which typically takes 3 to 10 minutes. You'll receive a notification when your avatar is ready.
• Review and confirm: When processing is complete, Google shows you a preview of your avatar in a short generated clip so you can evaluate how well it captures your appearance and voice.

Why Does Character Consistency Matter for Creators?

One of the hardest problems in AI video creation has always been keeping the same character looking the same from scene to scene. With your personal avatar saved as an asset in Flow AI, you always look like you across every clip and every project. This solves a critical workflow problem for creators who want to produce multi-scene videos or series featuring themselves without the logistical challenges of filming, lighting, and camera setup.

The avatar feature integrates directly into Flow AI's existing creative workflow, meaning you can use your avatar the same way you use the Ingredients to Video system and all the other tools in the platform. This integration ensures consistency not just in appearance, but in how the avatar functions within your broader creative projects.

What Should Creators Know About Quality and Limitations?

The quality of your avatar depends heavily on the conditions during setup. Poor lighting during the face scan or background noise during voice recording are the most common causes of suboptimal results. Most creators find their first attempt is good enough to work with, and the system allows you to redo the scan if the avatar doesn't look quite right.

The rollout is still ongoing. If you don't see the avatar feature in your Flow AI account yet, your account may still be in the rollout queue. Google is expanding access gradually, so checking back in a day or two may reveal the feature. This phased approach allows Google to manage server load and gather feedback from early users before full-scale deployment.

The introduction of Gemini Avatar represents a significant shift in how creators can produce video content. By removing the need for cameras, lighting setups, and location scouting, the technology democratizes cinematic video production. Creators can now generate videos of themselves in any setting, from a beach at sunset to a fantasy forest to a professional stage, without leaving their workspace. This capability opens new possibilities for content creators, educators, marketers, and anyone who wants to produce professional-quality video content featuring themselves as the main character.