Google did not really launch a video toy this week.
On Nov. 5, Google DeepMind put Gemini Omni 1.2 Flash into general availability on the Gemini API and Gemini for business, and Google Cloud repeated the news on Nov. 6.
The general availability is the first time a buyer can treat Omni like infrastructure instead of a demo product.
Matt Rouif, CEO and co-founder of Photoroom, said in the Google Cloud post:
“Photoroom turns product photos into video ads that sell in minutes. We chose Gemini Omni Flash to power our Video Ad Maker and Product Animator because it balances quality, speed and cost. Version 1.2 brings 15-second videos, a top user request. With general availability and SLAs, we can rely on it in production for our sellers and enterprise customers.”
I did some math to help you understand this in a more meaningful way.
At 720p, a second costs $0.07. That is 30% less than 1.1. A full 15-second clip lands at $1.05 if you use the rounded rate, $1.04 if you use the token math Google actually bills by: $12 per million video output tokens, 5,792 tokens for each second of 720p.
Three cents a second doesn’t sound like much. Right?
But, across 500 product pages, it starts to look like a budget line, not a toy.
General availability is where the money changes shape.
Google Cloud blog post said the model runs on enterprise infrastructure and security, with Provisioned Throughput available for high-volume work. Outputs carry an imperceptible SynthID watermark. That is the difference between a model you test and a model you put in a contract.
Like seriously?
We are paying real dollars to get a video with a watermark that we can’t remove!
Well, the workflow is the other price cut.
Draft at 360p or 720p. Edit by chat. Upscale only the one you keep to 1080p or 4K. Before, a wrong background meant starting over. Now it means a follow-up prompt. Swap the product. Change the light. Change the weather. Change the aspect ratio. The clip stays and here’s when paying for this feels good.
Fifteen seconds, up from ten. Three to fifteen, set in one-second steps. That is long enough for a full product explainer or a social spot without stitching clips together that would take hours of manual work.
The model will take up to six reference images, plus short motion clips up to three seconds and audio cues, and try to hold faces and packaging type steady across angles. Brand colors get the same treatment. Audio references are still gated behind allowlisting. Not everyone gets them yet.
Speech got less robotic and lip-sync is tighter. Voices stay attached to the right character when two people talk.
Camera moves are calmer too. A flythrough or an orbit does not jump as often. I’ve seen enough launches to know the demo always looks clean.
The test is whether that hold survives a catalog like when we are going to create an actual short drama with this updated model, not a single poster.
You can find it in Google AI Studio and in the Cloud console as it’s available there now. It is also in Flow and Google Vids.
Ultra subscribers get it in the Gemini app. Google Ads gets it too, so you can now create up to 15-second video ads for around $1 (that sounds great, but it’s AI, not original).
In Vids, 15-second clips come at no additional cost with a Google account or Workspace subscription.
Google said rollout to Rapid Release and Scheduled Release domains would take one to three days from Nov. 5.
Shivam Rastogi, VP of Engineering at invideo, said:
“We ran Gemini Omni 1.2 Flash through our evals at invideo. Honestly, it’s really good. It sticks to the prompt, and the physics and sound feel real. If you need fast video at scale, it’s one of the best we’ve tested. And 15 seconds means a full scene in one go.”
Google is not pretending the hard parts are solved. Fast sports action is still listed as work in progress. So are rapid script changes and complex physics.
The company published no quantitative benchmarks in the two announcement posts, only side-by-side examples.
So before you go for it and trust the demo reel, you should wait for the actual tests or examples.
All in, you can now use Gemini Omni 1.2 Flash to turn photos you already have into a 15-second ad, fix one detail by chat, and pay to upscale only the version you keep. You don’t need a shoot for every variant.
The Vids rollout was supposed to take one to three days from Nov. 5. By the end of this week, we will know if the consistency holds when thousands of sellers try it at once.
