← all releases
Google DeepMind·May 19, 2026·Gemini Omni v1·ai_model
Gemini Omni
Gemini Omni is Google DeepMind's new flagship multimodal AI model announced at Google I/O 2026. It can create anything from any input — starting with video — combining an intuitive understanding of physics with Gemini's real-world knowledge for natural-language conversational editing of video, images, and more.
featured on blogMultimodal AIVideo GenerationVideo EditingGoogle DeepMindGeminiGenerative AIGoogle IO 2026
on the blogGemini Omni: Google DeepMind's Multimodal Model That Creates Anything From Anything 4 min
overview
Gemini Omni is Google DeepMind's new flagship multimodal AI model announced at Google I/O 2026. It can create anything from any input — starting with video — combining an intuitive understanding of physics with Gemini's real-world knowledge for natural-language conversational editing of video, images, and more.
key features
- 01Create anything from any input — text, image, video, audio all supported as input modalities; generates and edits video with deep world-physics understanding and conversational language control
- 02Conversational editing pipeline — edit characters, backgrounds, scenes, and other video elements using natural language voice or text commands without timeline-based editing tools
- 03Integrated into Gemini app and Google Flow — available via Google AI subscriptions (AI Plus, Pro, Ultra) and developer API; powers Google Flow creative AI platform
related productsbrowse all →