5人のクリエイターが「Gemini Omni」で作ったものを紹介See what 5 builders are making with Gemini Omni

Google has spotlighted five creators and the projects they built with Gemini Omni, a model the company describes as making video creation "as easy as having a conversation." Published on Google's Keyword blog, the collection matters because it shifts the discussion around generative AI away from benchmark claims and toward concrete, everyday use, showing how editing footage and visualizing ideas can be handled through natural dialogue rather than specialized software skills.

The central premise is that Gemini Omni lets people describe what they want in plain language and see it reflected in a video. Instead of navigating multi-track timelines, keyframes and effects panels, users appear to be able to request changes conversationally, such as trimming clips, restructuring sequences, or turning a rough concept into a visual draft. Google frames the five examples as evidence of real-world creative applications, suggesting the emphasis is less on technical novelty and more on lowering the barrier to entry for people who are not trained editors.

The "Omni" name signals the model's multimodal scope. Omnimodal systems are designed to take in and produce more than one type of data, including text, images, audio and video, within a single model rather than by stitching together separate tools. That approach is central to why conversational editing is feasible at all: the system needs to understand a spoken or typed request, interpret the contents of a video, and generate or modify visual output in response. Google has been building toward this with the broader Gemini family, which has emphasized native multimodality since its earlier releases.

For readers unfamiliar with the landscape, Gemini Omni sits alongside a growing set of Google generative-media tools. The company has developed Veo for video generation and Imagen for still images, along with experimental filmmaking interfaces intended to help storytellers assemble scenes with AI assistance. Positioning a conversational editing model within that ecosystem is consistent with Google's stated goal of integrating generative capabilities across its products, from the Gemini app to Workspace and cloud services aimed at developers and enterprises.

The move also reflects wider industry momentum. Generative video has become one of the most competitive areas in AI, with OpenAI's Sora, Runway, Pika and Adobe's Firefly-based video features all pursuing similar territory. The naming convention itself echoes a trend, as OpenAI used "omni" for GPT-4o to denote a single model handling multiple modalities, and the marketing focus on ease of use mirrors how many vendors now pitch their tools to non-specialists rather than only to professional post-production teams.

Framing the announcement around five builders is a familiar format for Google, which frequently profiles individual users to illustrate how a product performs outside a lab setting. The approach is useful because it grounds abstract capabilities in specific outcomes, though it is worth noting that curated case studies are chosen to show a tool at its best and may not reflect the full range of results a typical user would achieve. Based on the excerpt, the blog post does not appear to detail pricing, availability by region, or the specific technical limits of the model.

Several practical questions remain open. It is not clear from the summary how much of the video work is generated from scratch versus edited from user-supplied footage, how long clips can be, or what safeguards Google applies around copyrighted material, likenesses and synthetic media disclosure, all issues that have drawn scrutiny across the generative-video sector. Watermarking and provenance tools, such as content credentials and Google's own SynthID approach to labeling AI-generated media, are increasingly relevant as these systems reach mainstream creators.

For now, the takeaway is narrower but still meaningful: Google is presenting Gemini Omni as a conversational front end to video creation and idea visualization, and it is using real creators to demonstrate what that looks like in practice. Whether the conversational model becomes a mainstay of everyday editing will likely depend on how reliably it handles complex, iterative requests and how it is ultimately priced and rolled out to the wider Gemini user base.