- Google has officially announced “Gemini Omni,” a new family of AI models capable of high-quality video generation and editing by combining multimodal inputs such as images, audio, video, and text at “Google I/O 2026.”
- “Gemini Omni” is a new family of AI models that allows for easier video generation and editing through natural conversation.
- The first model, “Gemini Omni Flash,” is being rolled out to “Google AI Plus,” “Google AI Pro,” and “Google AI Ultra” members via the Gemini app and “Flow.”
On Tuesday, May 19, 2026 (local time), Google officially announced “Gemini Omni,” a new family of AI models capable of not only high-quality video generation using multimodal inputs like images, audio, video, and text, but also video editing, at the developer event “Google I/O 2026.”
“Gemini Omni” is a new family of AI models designed to make video generation and editing easier through natural conversation. Users can issue instructions to change elements within a video or add new objects while maintaining character consistency and scene context, all through natural dialogue.
Additionally, the feature includes the ability to generate a digital avatar by training the model on the user’s own appearance and voice, allowing them to include the avatar in their videos. With ready-to-use templates provided, even users without professional equipment or technical skills can easily generate and edit videos.
Videos generated with “Gemini Omni” will incorporate “SynthID,” a digital watermarking technology that is invisible to the human eye.
The first model in the “Gemini Omni” family, “Gemini Omni Flash,” is being deployed for “Google AI Plus,” “Google AI Pro,” and “Google AI Ultra” members via the Gemini app and “Flow.” It is also scheduled to become available for free on “YouTube Shorts.”
Source: Google (https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-omni/)
コメントを残す