New Delhi: Google announced Gemini Omni, a new family of AI models that combines inference with video generation and editing, marking the company’s next step in multimodal AI creation tools.
The company launched the first model in the family, Gemini Omni Flash, with the Gemini app, Google Flow, and YouTube Shorts. The tool is rolling out to YouTube Create users this week.
Google says Gemini Omni allows users to generate and edit videos using natural language prompts. Users can upload text, image, audio, and video references to create clips, while the model maintains scene continuity, character consistency, and realistic physics throughout the edit.
One of the key features highlighted in this blog is conversational video editing. Users can adjust videos across multiple prompts, changing environments, styles, camera angles, and objects without losing the context of the original scene. Google says the model can also change actions in the video, such as transforming objects or adding new visual effects with prompts.
The company says Gemini Omni is designed to go beyond photorealistic visuals by combining Gemini’s reasoning abilities with an understanding of physics, history, science, and cultural context. According to Google, the model provides a better understanding of motion, gravity, and fluid mechanics, which helps create more realistic video sequences.
Google also introduced Omni’s ability to generate explainer-style videos and educational content from short prompts, including claymation-style scientific explanations and fast-paced animated storytelling formats.
This model supports multimodal input, allowing creators to combine references from images, video, text, and audio into a single generated output. Voice references will be supported on the audio input at launch, with broader audio support expected later.
Google is also introducing AI-generated digital avatars that allow users to create videos using their likeness and voice. The company said it is taking a cautious approach to broader audio and audio editing capabilities as it continues to test responsible implementation practices.
All videos generated through Gemini Omni include Google’s SynthID watermarking technology for AI content identification and verification.
Gemini Omni Flash is rolling out globally to Google AI Plus, Pro, and Ultra subscribers through the Gemini app and Google Flow. Google says this model will be made available to developers and enterprise customers through an API in the coming weeks.
