Google and several employees highlighted a set of Gemini 4 Argon demonstrations on Oct. 13, showing how creators combined the model with Google Antigravity, Blender and browser-based 3D tools. The examples cover visual storytelling, games and a simulated robotics project, but they are company-selected demos rather than independent performance tests.
Bees, robots and 3D worlds
In one example, Google said a creator used Argon’s 3D modeling and spatial reasoning in Antigravity to visualize “the daily life of a bee”. Another Google post described a Rubik’s cube-solving robot built end to end in Antigravity. The project modeled the robot arms in Blender, ran a real-time simulation in Three.js and used the Interactions API to solve the cube.
Google also highlighted a butterfly museum rendered from scratch and complex visuals created entirely from code in Google AI Studio.
Games and an animated Pixel
A Google developer-relations employee said she used Argon with Blender MCP and Three.js to turn photos of her home-office desk into a playable racing game. A separate creator showed Argon Tennis, a game that uses a phone as a tennis racket.
Sundar Pichai shared another Google-made example that started with a raw Pixel CAD file in Antigravity. He said Argon produced an exploded-view animation, including lighting, camera choreography and audio through Blender MCP, while running 1,000 iterative loops to check renders and refine code.
The posts do not detail how repeatable the results are, how much human editing each project required or what access users need to recreate them.
