World models & embodied AI

Systems that simulate environments
and reason in the physical world

Genie 3

A new frontier for world models

Gemini Robotics

Our most advanced vision-language-action model

SynthID

Embedding watermarks to identify content generated through AI


Experiments and prototypes

Models and experiments built with Gemini

Gemini Diffusion

Our diffusion architecture Gemini models

Project Mariner

Exploring capabilities for the universal AI assistant

Project Astra

Exploring the future of human-agent interaction

Model cards

Simple, structured overviews of how an advanced AI model was designed and evaluated.


Start building

Google Antigravity

Experience an agentic development platform, evolving the IDE into the agent-first era

Vertex AI Studio

Explore 200+ models on our enterprise platform with tools and features for AI development


Latest news

Project Genie: Experimenting with infinite, interactive worlds

Models

Veo 3.1 Ingredients to Video: More consistency, creativity and control

Models

Gemini 3 Flash: frontier intelligence built for speed

Models

Improved Gemini audio models for powerful voice experiences

Models

Start building with Gemini 3

Models

Build with Nano Banana Pro, our Gemini 3 Pro Image model

Models