Gemini Omni 1.2 Flash: Fast and cost-effective with longer clips

AI video is moving from clip toys to production plumbing. Gemini Omni 1.2 Flash is generally available on the Gemini API and Google Cloud with native 15-second generation — up from 10 seconds — duration control from 3 to 15 seconds in one-second steps, and conversational multi-turn editing so you tweak a shot instead of regenerating from scratch.

Price drops 30% from Omni 1.1 to $0.07 per second for 720p. Google also pushes stronger reference adherence for characters, products, and style, plus clearer multi-speaker speech and steadier cinematic camera moves. Outputs carry SynthID; provisioned throughput is available.

Cheaper, longer, editable clips change who can afford iterative video ads and explainers.

We're treating this as Google industrializing Omni rather than debuting a magic reel.

The workflow pitch is sane: draft at 360p or 720p, fix with follow-ups, upscale winners to 1080p or 4K. Virtual try-ons that keep garment drape across a 15-second spin, poster builds that lock typography, and unbroken FPV moves through architecture are the demos that matter for commerce and brand teams. Multi-turn background swaps that preserve performances are how you make ad variants without a new brief each time.

Google is candid that fast sports, rapid script changes, and complex physics still need work. Good — that honesty belongs next to the generally-available stamp. Useful generative video earns its seat when consistency and cost let teams ship weekly creative, not when a single hero clip goes viral. The downside if SynthID and provenance stay invisible to platforms: a flood of polished synthetic ads and explainers that audiences cannot audit. Price and length help builders; watermarking and disclosure have to keep up.

Omni lands across AI Studio, Cloud console, Flow, Google Vids, the Gemini app for Ultra subscribers, and Google Ads. It sits in a market where several labs are racing on seconds of coherent video and brand-safe reference control.