AI video just got easier to fix. Google made Gemini Omni 1.2 Flash generally available on November 5, and the big change is not just longer clips — you can edit a video by chatting with it, instead of starting over every time one detail is wrong.
What changed
- Longer clips: 3 to 15 seconds, in one-second steps — up from a 10-second limit before.
- Conversational editing: ask for a change over several turns, like swapping a background, and keep the rest of the shot.
- Use your own references: feed it an image, a short motion clip, or audio to help keep a person, product, or style consistent.
- Cheaper 720p: $0.07 per second, which Google says is 30% less than the last version. A full 15-second clip is about $1.05 in generation cost.
Try it like this
- Start with a short, clear prompt: who or what is in the shot, what happens, and where it happens.
- Generate a low-resolution draft first, check it, then upscale the version you like — drafts are cheaper to throw away.
- Edit in chat instead of re-prompting from scratch: "keep the same character, change the background to a garage."
- Check the final clip yourself before you post it — faces, text, and hands can still come out wrong.
Good to know
- It is available through the Gemini API and Google’s business tools, so most people will meet it inside apps and services built on Gemini, not as a standalone button.
- Generated clips carry an invisible SynthID watermark, so the video is marked as AI-made even if you cannot see it.
Source: SXF Brief — AI News Today, Issue 007, Nov. 7, 2026, reporting Google’s Nov. 5 launch.
The tip: stop re-rolling whole videos. Draft cheap, edit by chatting, and only pay to upscale the take you actually want.
