Gemini Omni 1.2 Flash: 15-second videos and a 30% lower price

Google made its video model generally available. It handles clips up to 15 seconds, conversational edits and more consistent characters, at $0.07 per second for 720p.
What changed
Gemini Omni 1.2 Flash is now generally available through the Gemini API, AI Studio, Google Cloud Console, Flow, Google Vids, the Gemini app for AI Ultra subscribers and Google Ads.
Clips can run from 3 to 15 seconds, up from a maximum of 10. The model accepts character images, product photos, motion clips up to three seconds and audio cues as references, and is better at keeping faces, clothing, packaging typography and brand colours consistent across camera angles. It also handles continuous camera moves such as FPV drone flights with fewer unwanted jump cuts. Every output carries an imperceptible SynthID watermark.
Conversational Editing lets users change the background, lighting or aspect ratio without regenerating the whole clip. A tiered workflow drafts at 360p or 720p, refines details and upscales only the final result to 1080p or 4K.
Price and practical impact
The price is $0.07 per second for 720p video, 30% less than Omni 1.1 Flash, with similar savings at other resolutions. A 15-second clip costs about $1.05, a 10-second clip about $0.70.
That changes prototyping economics for agencies and small e-shops: a dozen variants of a product clip cost tens of dollars, not thousands. Final production still needs a human editor, especially for brand consistency and legal review.
In the EU, synthetic video must be disclosed under the AI Act's transparency rules. SynthID helps platforms detect it, but the obligation to label stays with the publisher.
Sources
More news
Google releases Gemini 4 Carbon: a frontier model for cybersecurity
Google's strongest model leads in finding vulnerabilities and fixing them. Access is initially limited to a handful of cyber defenders — a public release will follow later.
Read articleOpenAI releases GPT-6.1 Terra: a cheap model for computer work
The new model at $1 per million input tokens understands the screen, clicks through apps, and writes naturally. It costs a tenth of Astra and handles tasks 9× cheaper than the competition.
Read article