Google Flow hits I/O with Gemini Omni

Google is finally closing the loop between video generation and editing by making the process "agentic" rather than just prompt-based.

• –Gemini Omni's natively multimodal architecture allows for "any-to-any" generation, making it possible to use sketches or audio as direct references for video creation.
• –Conversational editing solves the "black box" problem of AI video by allowing iterative refinement—changing camera angles or lighting without rerunning the entire generation.
• –The shift to a subscription-based "AI Studio" model (Google AI Plus/Pro/Ultra) signals Google's move away from experimentation toward a professional-grade creative suite.
• –Integration with YouTube Shorts and Create suggests a strategy to capture the mobile-first creator market before OpenAI's Sora becomes widely available.
• –Character consistency and "world understanding" (physics-aware reasoning) are the key technical differentiators aimed at cinematic production quality.