Google launches Gemini Omni Flash: any-to-video model with physics-consistent editing
Gemini Omni Flash takes any combination of text, image, audio, and video as input and produces video output with physics simulation, scene memory, and character consistency — accessible today to AI Plus, Pro, and Ultra subscribers via the Gemini app and Google Flow, and free on YouTube Shorts. The strategic move is Google collapsing the gap between reasoning and generation in a single model, with SynthID watermarking baked in from day one and a developer/enterprise API still incoming.
Source: blog.google ↗
Omni is our new model that can create anything from any input — starting with video.
Google
Why this matters
- → Video editing via natural language conversation eliminates the need for manual frame-by-frame tools, lowering barriers to professional-quality output.
- → Physics simulation and scene memory maintain consistency across edits, enabling complex multi-turn creative refinements without losing coherence.
- → SynthID watermarking built in from day one establishes provenance for synthetic video at scale, addressing authenticity concerns before widespread adoption.
Reasoning meets generation