Multimodal · NOVEMBER 7, 2026
Gemini Omni 1.2 Flash adds 15-second video and cheaper ad iteration
Google cuts its stated 720p generation price by 30% and adds conversational revisions, but better creative economics are not evidence of better ad conversions.
Google launched Gemini Omni 1.2 Flash on November 5, adding video generation up to 15 seconds and pricing 720p output at $0.07 per second.
Google Cloud says that is 30% below Omni 1.1 Flash. That comparison is Google Cloud’s, against Gemini Omni 1.1 Flash. It’s a cost claim, not evidence of improved ad conversions.
Longer clips and revisable output
Google’s launch materials state: The release also adds conversational revisions and stronger reference adherence, capabilities aimed at product videos, virtual try-ons, and ads.
Google’s launch post says durations now run from 3 to 15 seconds in one-second increments, compared with a previous maximum of 10 seconds.
The post outlines the production sequence: It describes a workflow of drafting at 360p or 720p, revising details, and upscaling selected output to 1080p or 4K.
Google describes the controls and its fidelity claims:
Google says follow-up prompts can swap a product or background, change lighting, or switch aspect ratio without restarting the clip.
It also claims better preservation of packaging typography, garment details, and brand colors across camera angles.
Those claims aren't independent measurements of product accuracy.
Inputs, restrictions, and human review
Google DeepMind’s Gemini Omni 1.2 Flash Model Card describes multimodal support: The model card lists text, images, audio, and video as inputs, with video and audio as output.
Access isn't unrestricted. Google Cloud explicitly says audio references require allowlisting; separately, the model card says changing people’s speech is currently restricted.
Google also acknowledges reliability limits: Google itself says fast sports action, rapid script changes, and complex physics remain areas under development.
"The practical implication: cheaper revisions could expand creative testing, but product fidelity still needs human review."
For founder-led businesses, that suggests more room to test promotional variants from existing imagery. It doesn't establish that those variants attract or convert more customers, even with 15-second output.
Availability differs by route
Google Cloud says the model is generally available through its platform.
For Google Vids, the timing is more specific: Google Workspace’s more specific Vids notice says rollout began November 5, with feature visibility taking 1–3 days.
The same notice describes eligibility: Its availability list includes personal Google accounts and Business Starter, Standard, and Plus, among other editions; expanded-access licenses have higher usage limits.
The announcements differ on Gemini for business. The main launch post describes general availability across that product and the Gemini API, while the Cloud post says Gemini for business support is coming soon.
Vercel offers another integration route: Vercel says AI Gateway exposes google/gemini-omni-1.2-flash, supports up to six reference images, and allows follow-up revisions using previousInteractionId.
For an operator choosing a deployment route, the distinction matters: general availability in one channel doesn't settle access in another. Lower generation costs can support experimentation; they don't replace checking product fidelity or measuring campaign results.