Gemini Omni 1.2 Flash ships 15-second audio-video clips at about 30% lower cost

Omni 1.2 Flash is Google's cost-optimized multimodal generator. It produces native audio and video together, so sound and picture are synchronized, at 720p and up to 15 seconds per clip. Users can refine a clip over multiple turns rather than re-generating from scratch.

Pricing is the headline for developers. Output now costs roughly $0.07 per second of video, or about $12 per million tokens, about 30% below the prior $0.10 per second. In Google Vids, the model brings better character consistency and vocal realism to Workspace users. Google Forms also gained Gemini-powered insights in the same Workspace update cycle.

The launch follows Gemini 4 Carbon, which Koray Kavukcuoglu calls Google's strongest model. It positions Google against OpenAI's video efforts, xAI's Grok Imagine and a growing set of open video models in the Stable Diffusion community. Developers on dev.to welcomed the price cut, and InVideo said the physics and sound "feel real."

Fifteen seconds is still short for narrative work, and 720p trails premium competitors on resolution. The trust backdrop is also shaky: a "Gemini trains on chat history, even for subscribers" thread hit 715 points on HN this week. Watch for longer clips, higher resolutions and enterprise data-use guarantees.