AI video prompts for Gemini Omni Flash, Google's video model that generates matching audio along with the video. Create video from text or a photo, and change only the part you name in a clip you already have.
2026.08.27
Gemini Omni Flash is Google's video generation model. When it renders video from text it generates matching audio in the same pass, and it draws on Gemini's real-world knowledge and grasp of physics to show how objects move and interact. The latest version is 1.1. In Carat's model picker it appears as Flash under the Google Omni brand. Carat pairs each Gemini Omni Flash result with the prompt behind it, so you can copy and run it as is.
Gemini Omni Flash generates video and audio together rather than rendering them separately and stitching them, so dialogue and sound effects are prompt instructions. Text-to-video, image-to-video, reference-to-video and video editing all live in the same model, so building a shot and then revising it stay in the same prompt workflow. You can pick 360p, 720p, 1080p or 4K, and credit use follows the resolution and the length. Input formats for editing are mp4, mov, webm, m4v and gif.