BUSINESSNew tool · · 2 min read · Source: Google (The Keyword)
Gemini Omni 1.2 Flash: Google's AI video now runs 15 seconds
On November 5, 2026, Google introduced Gemini Omni 1.2 Flash, the new version of its AI video model. Clips can now run up to 15 seconds, can be edited turn by turn in a conversation, and the API price drops by about 30%.
My tool tip · AdGranola – AI meeting notesTry it free
Key facts
- Google introduced Gemini Omni 1.2 Flash on November 5, 2026; clips run 3 to 15 seconds, up from a 10-second maximum.
- In the Gemini API, 720p video costs $0.07 per second, 30% less than Omni 1.1.
- Available in the Gemini API, AI Studio and Vertex AI; in the Gemini app first for Google AI Ultra subscribers.
What's new
Omni 1.1 topped out at 10 seconds. With 1.2 Flash you set the length in one-second steps from 3 to 15 seconds. Google's main improvements:
- Conversational editing: change a finished clip with a follow-up instruction, such as swapping the product or background, without regenerating everything.
- Reference adherence: character, product and style images, motion clips up to 3 seconds and audio cues are meant to carry through more consistently.
- Speech: more natural dialogue, cleaner lip-sync and multiple speakers.
- Workflow: draft cheaply at 360p or 720p, then upscale to 1080p or 4K.
Google also lists limits: fast sports action, complex physics and rapid script changes are still hard.
Price and availability
In the Gemini API, 720p video costs $0.07 per second, 30% less than Omni 1.1. A 15-second clip comes to about $1.05. The model is generally available in the Gemini API, Google AI Studio and, for businesses, Vertex AI. In the Gemini app, 15-second videos roll out first to Google AI Ultra subscribers. It also powers Flow, Google Vids, Gemini for business and Google Ads, and Google says 15-second clips in Vids come at no extra cost. The announcement doesn't mention specific countries or the EU.
Labeling
Google says every generated video carries an invisible SynthID watermark that can later be used to check for AI origin.
What this means for you
- 15 seconds covers many Reels, Shorts and ads, making AI video more practical day to day.
- Editing by chat saves time and money because not every tweak needs a full regeneration.
- At roughly $1 per 15-second 720p clip, testing several variants is cheap.
How to use this
If you plan social clips or ads, write three short opening variants for one concept, generate them as cheap drafts and upscale only the best. Label the final video as AI-generated.
Luigi's take
For marketers, the key upgrade isn't length, it's conversational editing. Anyone who works with AI video knows the problem: a clip is almost right, but fixing one detail means rolling the dice again. If targeted edits now work, AI video becomes more predictable, and that's exactly what teams need when they ship ads or social clips on a schedule.
My advice: use the cheap draft tier for variants and test hooks before you upscale. And label AI video clearly, the way I label AI images on this website. An invisible watermark doesn't replace a visible, honest note for your audience.
Source
This article was researched and written with AI assistance. Responsible: Luigi Lugmayr.
DealTypingMind Black Friday: why the 50% off isn't a real discount
NewsMeta: Brazilian court blocks AI training on media content
NewsPerplexity Teammates: AI agents that stay on your projects