Runway adds Google's Gemini Omni Flash video model
By Steven Van ·
Creators can now generate and edit clips with conversational, physics-aware editing alongside Runway's own Gen-4.5.
Runway just wired Google's newest AI video model straight into its creative suite. As of July 1, 2026, you can generate and edit video with Gemini Omni Flash inside Runway — starting from a text prompt, an image, or an existing video clip and shaping it into whatever you can imagine. You can jump in directly or simply ask Runway's Agent to use Omni for you.
What Gemini Omni Flash brings to the table
Gemini Omni Flash is Google DeepMind's multimodal video model, first shown to consumers at Google I/O on May 19, 2026 and opened up to developers via the Gemini API and Google AI Studio on June 30. It takes text, images, and video as inputs and produces short 720p clips — durations run from 3 to 10 seconds — with the option of natively generated audio that matches the scene. The headline trick is that it treats generation and editing as a single conversation rather than two separate workflows.
Editing video by talking to it
The feature that sets Omni Flash apart from other API-accessible video models is conversational editing. Instead of regenerating a clip from scratch every time you want a change, you keep a chat session open and pass follow-up instructions as new turns. The model tracks context across those turns, so character consistency, lighting, and scene continuity carry over from one edit to the next. In practice, that means you can say "make it night," then "add rain," then "pull the camera back" and watch the same shot evolve — no re-prompting the whole scene each time.
Why this pairing matters for Runway users
Runway already ships its own state-of-the-art Gen-4.5 model, so adding Omni Flash gives creators a second engine to reach for depending on the job. Omni Flash's conversational, physics-aware editing complements Runway's existing toolkit — image-to-video, inpainting, motion tracking, green screen removal, and its Aleph editing tools — inside one web-based editor. And because Runway exposes it through the Agent, you don't have to learn a new interface: describe the change in plain language and let the Agent route it to Omni. Pricing on the underlying model lands at roughly $0.10 per second of 720p output, so a full 10-second clip costs around a dollar.
Sources: Google blog, VentureBeat, Runway API changelog.