Summary
Skill Leap AI presents twelve hands-on demos of Gemini 4 Argon, Google’s newly released model, based on several days of early access. The video highlights how the model handles 3D worlds, physics, and one-prompt game creation, along with more practical work like presentations and planning.
Demos include turning a Van Gogh painting into a walkable 3D room, building a solvable 3D Rubik’s Cube, converting a floor plan into a navigable house, turning a sketch into a playable platformer, and modeling roof load in an interactive 3D tool. A city-demolition physics demo shows where the model still struggles, with floating bricks and unconvincing collisions. The presenter also has Gemini plan a party and produce a Google Sheet comparing venues, pointing to the model’s native integration with Google products.
Practical notes include that Gemini 4 Argon appears in the model picker with low, medium, and high thinking levels, is available to Google AI Pro and Ultra subscribers, is offered through the Gemini API with introductory pricing, and works in Google’s Antigravity coding platform. The presenter also explains that simple prompts sometimes trigger Nano Banana image generation, and that asking for a Canvas interactive app fixes it. A new Tasks section and Gemini skills are also covered.
📺 Source: Skill Leap AI · Published October 13, 2026
🏷️ Format: Showcase
