Google's Gemini Omni 1.2 Flash: AI video is now generally available and 30% cheaper

Google announced Gemini Omni 1.2 Flash on 5 November 2026, an updated model for generating and editing video. It is the first version of Gemini Omni that Google offers as generally available, meaning it is out of preview and supported for production use through the Gemini API and Google Cloud.

What it does

Omni Flash turns a written prompt, plus optional reference material, into a short video clip with sound. Compared with version 1.1, Google lists these changes:

• Longer clips. Up to 15 seconds, up from 10. Length can be set anywhere from 3 to 15 seconds in one-second steps.
• Better use of references. You can supply images of a character, product or style, short motion clips of up to 3 seconds, and audio cues, and the model keeps those details steady through the clip.
• Less drift. Fewer cases where faces or objects morph between the first and last frame, including during camera moves and when a clip is extended.
• Edit instead of regenerate. You can refine a clip turn by turn, for example swapping the product or background, changing the lighting or switching the aspect ratio, without starting over. This works on full 15-second clips.
• Speech. More natural dialogue, clearer lip-sync and support for several speakers.
• Draft cheap, finish sharp. Drafts can be made at 360p or 720p, and finished clips upscaled to 1080p or 4K.

What it costs and where to get it

Google says pricing is 30% lower than version 1.1, at $0.07 per second of 720p video. A 15-second clip at that setting works out to about $1.05 before any upscaling. Check the Gemini API pricing page for other resolutions before you budget.

• Developers: the Gemini API in Google AI Studio, and the Google Cloud console, where reserved capacity (Provisioned Throughput) is offered for production use.
• Google Vids: 15-second clips at no extra cost with a Google account or Workspace subscription.
• Other Google products: Flow in Google Labs, the Gemini app (rolling out to Google AI Ultra subscribers) and Asset Studio in Google Ads for video campaigns.

Who should care

• Marketing teams producing many short variants of the same ad or product clip, where turn-by-turn editing saves regenerating each version.
• E-commerce and product teams who want consistent product shots across clips using reference images.
• Teams already on Workspace, who can try it in Google Vids without new spend.

Limits and caveats

• Known weak spots. Google says it is still improving fast sports action, rapid script changes and complex physics. Expect retries for those.
• Labelling. Every output carries SynthID, Google's invisible watermark, which tools can use to identify the video as AI-generated. Plan to disclose AI-made content where your market or platform requires it.
• Audio references. Google Cloud says using audio as a reference needs allowlisting, so not every account has it yet.
• Regions. The announcement doesn't list regional availability, so confirm it for your cloud region.
• Rights and likeness. Reference images of real people, logos or third-party products bring the usual consent and trademark questions. The model doesn't settle them for you.

What to do this week

• Pick one recurring video task, such as a product clip or social ad, and produce it in Google Vids or AI Studio at 720p. Time how long it takes against your current process.
• Test the edit loop: change one element at a time and see how much of the clip stays stable.
• Work out your monthly cost from seconds of finished video, not number of prompts, and include the drafts you throw away.
• Write a short internal rule for when AI video needs human review and a disclosure label.

The DevDrift helps teams build AI into their content and operations workflows, with a person approving what goes out.

Sources

• Google: Introducing Gemini Omni 1.2 Flash
• Google Cloud: Gemini Omni 1.2 Flash now generally available