Google releases Gemini Omni Flash video

Google began rolling out Gemini Omni Flash on May 19, adding a new video-generation and editing model to the Gemini app, Google Flow and YouTube Shorts. The company said Omni Flash is the first release in a new Omni model family built to create from mixed inputs, including images, audio, video and text. Google also said the model supports step-by-step editing through natural-language prompts rather than one-shot generation. ### What exactly did Google ship? Google said the new product is Gemini Omni Flash, the first model in its Omni family. In a company post, Google described Omni as a system that can “create anything from any input — starting with video,” and said Omni Flash combines Gemini’s reasoning with Google’s generative media models. The DeepMind model card says Gemini Omni Flash is designed for high-quality video creation and conversational editing. (blog.google) Google’s prompt guide and product pages describe a workflow in which users can refine a clip across multiple turns, keeping a scene consistent while changing elements such as backgrounds, captions or other details. ### Where is Omni Flash showing up first? (blog.google) Google said Omni Flash is rolling out in three consumer surfaces first: the Gemini app, Google Flow and YouTube Shorts. Sundar Pichai’s I/O keynote post said the model was available starting May 19, while a separate roundup of I/O announcements said it was rolling out to Google AI Plus, Pro and Ultra subscribers globally through Gemini and Flow. (deepmind.google) YouTube Shorts Remix and the YouTube Create app are also part of the launch. Google’s I/O roundup said those YouTube tools are getting Omni Flash access for users age 18 and older at no cost. ### How does Google describe the editing workflow? Google said the model is built for iterative editing rather than a single prompt-and-output exchange. (blog.google) The company’s product materials compare Omni to “Nano Banana, but for video,” and say each edit builds on the last one so a scene stays coherent over multiple revisions. The Gemini app post said Omni can transform text, image and video prompts into “cinematic, high-quality video outputs.” The prompt guide adds that users can ask for specific updates without rewriting the full prompt, which Google presents as a conversational editing flow. (blog.google) ### What safety and provenance details did Google disclose? (blog.google) Google said videos created with Omni include SynthID, its digital watermarking system. In the I/O announcements post, the company said the watermark is imperceptible and that generated content can be checked through the Gemini app, Gemini in Chrome and Search. The DeepMind model card says the document covers intended uses, limitations, evaluations and content-safety considerations. (blog.google) Google did not present Omni Flash as a general-release developer API on launch day, instead separating the consumer rollout from later enterprise and API availability. ### What comes next after the app rollout? Google said more output types are planned beyond video. (blog.google) In its launch post, the company said image and audio outputs will be supported over time, expanding Omni beyond the initial video release. Google also said developers and enterprise customers will get access in the coming weeks. That next step, outlined in Pichai’s I/O keynote post, would move Omni Flash from consumer-facing apps into API and business use after the initial May 19 rollout. (deepmind.google) (blog.google 1) (blog.google 2)