Instead of a maximum of 10 seconds, a clip from Google's video model can now be 15 seconds long. Google introduced Omni 1.2 Flash on November 5, 2026 in the Google Blog. The model generates and edits videos and is available in the Gemini API, in Google AI Studio, in Google Cloud, in Flow, and in Google Vids.
What can Omni 1.2 Flash do better than Omni 1.1?
Google lets you choose the length in one-second steps from 3 to 15 seconds. As a template, the model combines images of characters, products, or styles with audio specifications and with clips of up to 3 seconds in length as a motion reference. According to Google, faces, clothing, and brand colors are supposed to stay consistent across camera angles.
Omni 1.2 Flash in numbers
Video model from Google, introduced on November 5, 2026
15 slongest clip, previously 10 seconds
30 %lower price per second than with Omni 1.1
3 slongest clip as a motion reference
3free videos per month in Google Vids
All values are statements from Google, no independent test is available yet
Sources: Google Blog, Google Workspace Blog, The Decoder
In the dialog, you can change the product, background, lighting, or aspect ratio step by step without having to start over. Also new are more natural speech, more precise lip sync, multiple speakers, and smoother camera movements with fewer cuts.
Google itself names weaknesses: for fast sports action, rapidly changing scenes, and complex physics, the team is still working on improvements. All the improvements mentioned are manufacturer claims; an independent test is not available in the sources read.
What does Omni 1.2 Flash cost, and what is free?
Google reduces the price per second by 30 percent compared with Omni 1.1. For 720p, the announcement gives the values in the table; for 1080p and 4K, Google does not state a price.
Clip at 720p
Price in US dollars
1 second
around 0.07
15 seconds
around 1.05
The total for 15 seconds comes from my own calculation. All amounts are in US dollars; a euro price for Germany is missing.
Google recommends a staged workflow: generate drafts in 360p or 720p, incorporate corrections in further editing steps, and only upscale the final version to 1080p or 4K. The Decoder sees this as an offering for developers and advertisers with high production volume.
Google Vids now also runs Omni 1.2 Flash. With a Google account or a Workspace subscription, you can generate up to three 15-second videos per month there at no extra charge, according to Google's Workspace Blog. Google names product ad clips in a setting without cuts, image-to-video with reference images, and more natural speaker voices as areas of use.
In the Gemini app, Google is initially enabling the model for subscribers of Google AI Ultra; the model is also available in Google Ads. Google gives no list of countries either in the announcement of the model or in the one for Vids, and it also does not mention a restriction to individual countries. Whether the feature already appears in your German interface can be shown by a test.
What this means for you
Feel free to try out Vids with a product photo as a reference image. Three free videos per month are used up quickly, so it's best to write out the prompt beforehand and then check whether the colors and shape of the product stay consistent over the 15 seconds.
If your project runs via the API, first calculate one clip at 720p and multiply the result by the number of versions your client should see. Only upscale the approved version.
Every video contains an invisible SynthID watermark. Viewers cannot see the watermark, so you have to add a visible label yourself if a client or a platform requires one. For client projects, clarify in advance whether AI-generated videos are desired.
For subjects with fast sports scenes or complicated physics, it's best to plan with a fallback, as long as Google is still working on it.
Google introduced Omni 1.2 Flash on November 5, 2026.
