Google this week launched Gemini Omni 1.2 Flash, the new version of its video generation model, and cut its price by about 30%. According to reports, generating 720p video now costs about 7 cents per second, and the model can make clips up to 15 seconds long, up from 10 in the previous version. It is available to developers through Google AI Studio and Vertex AI, and requests to the previous version will be routed automatically to the new one in early December.
To grasp what that means, do a simple sum: a 15-second clip costs a few tens of cents. Two or three years ago, video of this quality required a crew, equipment and days of work. Even if quality is not perfect yet, and Google itself notes fast sports motion and complex physics still need work, the direction is clear: synthetic video is becoming very cheap and accessible.
Who benefits? Advertising and marketing firms wanting many versions of an ad, creators on YouTube and social media, app developers building video into products, and small businesses that could not afford production. Competition is fierce: OpenAI and other video model makers improve their tools every few months, and prices are falling.
But there is another side. The cheaper and more convincing synthetic video gets, the easier it is to make scam and deepfake videos too: a celebrity recommending an investment, a 'CEO' asking for a transfer, or a fake news clip. The big companies' tools usually include digital watermarks and usage limits, but not all tools do. The advice to viewers is unchanged: when a video asks you for money, details or urgent action, check the source before believing your eyes.
