At this year's I/O Developer Conference, Google officially released the new generation of Gemini 3.5 series models and the new Gemini Omni series. Gemini 3.5 Flash has been the first to be opened to the public, while Gemini Omni focuses on the ability to "generate video from any input."
Gemini 3.5 Flash is already available to all users through the AI mode in the Gemini app and Google search. Google claims that the model has a level of intelligence comparable to large flagship models in many dimensions, while maintaining the consistent high-speed performance of the Flash series.
According to the introduction, Gemini 3.5 Flash is currently the strongest Gemini agent and code generation model. It even surpasses Gemini 3.1 Pro in benchmark tests related to complex code and agents, and maintains the lead in multi-modal understanding, so it is also set as the current default model. Users can directly call this model in daily searches, in-app conversations and other scenarios to get faster and smarter answers and code support.
Gemini Omni, unveiled at the same time, represents Google's latest attempt in the direction of generative video. This new model can generate videos from any input. Users can freely combine images, audio, videos and text as input, and the model generates high-quality video content based on Gemini's real-world knowledge. After the generation is completed, users can also perform multiple rounds of conversational editing through natural language to modify and fine-tune the details in the video.
The first model in the Gemini Omni series is Gemini Omni Flash, which supports partial or overall modification of the video and can continue to retain the narrative coherence of the original scene during multiple rounds of creation, allowing users to continuously adjust without losing the main story line. Google said that this model has a more intuitive understanding of physical concepts such as gravity, kinetic energy, and fluid dynamics, allowing it to generate more realistic and credible dynamic scenes.
In terms of creative experience, Gemini Omni allows users to use their own voices and avatars to participate in video creation, thereby generating digital avatars with personal characteristics. In order to deal with the traceability and security issues of synthetic content, all videos generated by this model will be embedded with SynthID digital watermarks for labeling and identifying AI-generated content.
In terms of availability, Gemini Omni Flash is available to global subscribers starting today. Users who subscribe to the Google AI Plus, Pro, and Ultra plans can use the model directly in the Gemini app and Google Flow. At the same time, Google is also introducing this capability to YouTube Shorts and YouTube Create for free, allowing more creators to experience the video generation function based on Gemini technology on mainstream content platforms.