Google has published eight examples of Gemini 4 Argon, its newest model, performing tasks involving research, coding, visual media and interactive 3D experiences.

According to Google, Argon generated 5,000 lines of mathematics from a short prompt to render a sunlit 3D library, then created a video about the process. The company also says the model produced an audiovisual tool that traces shapes across video frames and turns their movement into musical notes.

In another demonstration, Google used Argon with Antigravity, Blender MCP and Nano Banana to transform photographs of a physical desk into a playable 3D track. Google says Argon also processed more than 80,000 NASA files, including weather, engineering, health and science data from 72 Mars flights, and organized the information into an executive presentation using a supplied slide template.

Other examples include transforming scientific concepts from a PDF into an illustrated animation, generating a self-navigating 3D periodic table with synchronized audio, coding a custom fur engine for real-time hair dynamics, and creating an interactive musical object using fluid physics and reactive sound.

Google describes these examples as demonstrations of Argon’s reasoning, visual understanding, coding and tool-coordination capabilities. The post says Gemini 4 Argon is globally available through the API in AI Studio and the Gemini platform, in Google Antigravity for Google AI Ultra and Pro subscribers, and in the Gemini app and AI Mode in Search for those subscribers.