11 demos of Gemini Omni and Gemini 3.5 in action
Google's latest models bring conversational video editing and agentic workflows to the masses. Here's what they can do.
Condensed by AI-Portable from Editorial queue.
At Google I/O 2026, the tech giant pulled back the curtain on two new AI powerhouses: Gemini Omni and the Gemini 3.5 family, starting with 3.5 Flash. These models aren’t just incremental updates—they redefine what’s possible when reasoning meets creation and action. While the blog post walks through 11 flashy demos, the real story is how these capabilities are about to trickle into our pockets, apps, and wrists.
Gemini Omni is the showstopper for creatives. It takes any mix of images, audio, video, and text and generates high-quality video grounded in realistic world knowledge. But the magic isn’t just generation—it’s the conversational editing. You can shoot a clip of a violinist, then ask Omni to transport her into a different environment, tweak the camera angle over her shoulder, and even replace the instrument with something surreal—like a sculpture made of bubbles. Every prompt builds on the last, preserving character consistency, physics, and scene memory. One demo morphs a simple hand gesture into an infinite recursive room inside a floating glass sphere, while another reimagines a dinner-party video by dimming the lights and adding a checkerboard-globe effect. The takeaway? Your phone’s camera roll becomes a starting point for cinematic creations you could never film alone.
Gemini 3.5 Flash, on the other hand, is built for utility. It balances frontier intelligence with the speed you’d expect from a Flash model, making it ideal for agentic, long-horizon tasks. In one demo, powered by Google’s Antigravity harness, 3.5 Flash automatically renames and categorizes hundreds of unstructured assets based on dynamic criteria—no human intervention needed. Another showcases it spinning up 64 fractal variations in parallel, or prototyping multiple checkout-flow UX designs in 60 seconds on AI Studio. The model even visualizes complex concepts like Pi into interactive web UIs, all while coordinating sub-agents to tackle multi-step workflows under supervision.
But the most immediate impact hits closer to home. Gemini 3.5 Flash is now the default model for the Gemini app and AI Mode in Search globally. That means information agents running 24/7—background assistants that notify you the moment your favorite athlete drops a sneaker collaboration. Search itself is getting a generative UI overhaul: instead of links, you’ll get custom visual tools, dashboards, and interactive simulations built on the fly. And there’s Gemini Spark, a personal agent running on 3.5 Flash, deeply integrated with Workspace apps to handle tasks like finding nut-free snacks and adding them to your Instacart cart.
For portable AI enthusiasts, these demos aren’t just cloud curiosities. Omni’s video generation and editing will soon hit YouTube Shorts and mobile apps, while 3.5 Flash’s agentic smarts are already weaving into Wear OS 7’s proactive assistance. The models that can reason across text, code, images, and video are the same ones that will power your next smartwatch glance or hands-free voice command. Google is betting that the future of AI isn’t just answering questions—it’s acting on your behalf, continuously, whether you’re filming a home movie or tracking a cross-country move. The 11 demos are a teaser; the real test begins when these models land in our daily routines.