Gemini Omni is 'like Nano Banana for video' and I couldn't believe the results

What's New

Google appears to be pushing further into multimodal generative AI with a tool being described in early hands-on coverage as 'Gemini Omni' — a feature that reportedly lets users generate video from virtually any combination of inputs: text prompts, still images, audio clips, and even existing video footage. According to the reporting, this capability is showing up on Pixel devices, suggesting Google may be positioning it as a flagship feature tied closely to its own hardware ecosystem.

The comparison drawn in the coverage — likening it to 'Nano Banana for video' — references Google's earlier Nano Banana image generation model, which drew attention for its speed and quality in producing AI-generated images from prompts. If Gemini Omni delivers a similar leap for video, it would mark a meaningful step in closing the gap between text-to-image and text-to-video generation, historically a much harder technical problem due to the need for temporal consistency across frames.

Why This Matters

Video generation has lagged behind image generation in both quality and accessibility. Tools like OpenAI's Sora and various open-source models have shown flashes of impressive output, but consistent, user-friendly, multimodal video generation — accepting audio and existing video as inputs, not just text — has been rarer. If Gemini Omni truly allows blending of text, images, audio, and video into a single generative pipeline, it suggests Google is trying to differentiate its offering not just on output quality but on input flexibility.

This also matters strategically. Google has been racing against OpenAI, Runway, Meta, and others in the generative video space, and integrating such a tool directly into Pixel phones would be a distribution advantage few competitors can match. Baking advanced AI video tools into hardware that ships to millions of users could accelerate mainstream adoption of AI-generated video content, for better or worse — from creative filmmaking shortcuts to renewed concerns about synthetic media and misinformation.

The Bigger Picture

The framing as a mobile-first, on-device-adjacent experience (via Pixel) rather than a standalone web tool is notable. It hints that Google may be trying to make generative video as casual and accessible as taking a photo, mirroring how Nano Banana reportedly simplified image generation.

Until more technical details, availability specifics, and independent testing emerge, claims about output quality should be treated as early impressions rather than benchmarks. But if Gemini Omni performs as described, it could pressure competitors to accelerate their own multimodal video roadmaps, intensifying an already fast-moving race in AI video generation heading into next year.