Nine Demos Showcase Gemini Omni's Conversational Video Editing and Gemini 3.5 Flash's Agentic Workflows

Nine short demo videos posted May 29, 2026 put the new Gemini model family on display, showing how Gemini Omni and the Gemini 3.5 line aim to change creative and automated workflows. The clips present concrete examples of multimodal video generation and conversational editing with Omni, alongside agentic, action‑oriented tasks driven by the Gemini 3.5 family — led in these examples by 3.5 Flash — demonstrating both creative and production uses that builders can adopt.

Gemini Omni is shown accepting mixed inputs — images, audio, video and text-and both generating high‑quality video grounded in model knowledge and editing existing footage through conversational prompts. The demos emphasize iterative, instruction‑by‑instruction editing: successive prompts build on prior changes while keeping characters consistent, preserving physical continuity and maintaining scene memory so edits chain naturally rather than producing disjointed results.

The Gemini 3.5 family is presented as combining frontier intelligence with action; Google launches the series with 3.5 Flash, which targets agents and coding workloads. 3.5 Flash is described as delivering frontier performance on complex, long‑horizon tasks while remaining fast enough for productionized agentic flows and developer tooling. When paired with the updated Antigravity harness, 3. managed by collaborative subagents.

Developer‑facing integrations are a focal point of the demos. 3.5 Flash is used to generate multiple UX approaches for a checkout flow in about 60 seconds on AI Studio, and the model is now the default for the Gemini app and for AI Mode in Search globally. The material positions 3.5 Flash as building on Gemini 3’s multimodal foundation to power richer web UIs and interactive graphics, and illustrates how the updated Antigravity setup orchestrates subagents to preserve high performance in complex pipelines.

For builders and creators the practical implications are concrete: Omni lowers friction for iterative, prompt‑driven video production and scene editing using multimodal inputs, while 3.5 Flash together with Antigravity focuses on automating long, multi‑step workflows, asset management and rapid UI prototyping. The nine demos serve as reference material for integration patterns and for evaluating supervision, orchestration and performance tradeoffs when moving these capabilities toward production deployments.