What Is Gemini Omni 1.2 Flash? 15-Second AI Video, Editing, and the 30% Price Cut
Gemini Omni 1.2 Flash is Google’s newest AI video model. Announced on 5 November 2026, it generates and edits video from text prompts, images, video clips and audio — and it is the version of the model that Google is now offering as a general, production-ready service. The headline changes are easy to state: clips run up to 15 seconds instead of 10, you can edit a clip in conversation instead of starting over, and the price is 30% lower than the previous version.
This guide explains what Gemini Omni actually is, what changed in version 1.2, how to use it, what it costs with worked examples, who it suits, and where it still falls short.
What is Gemini Omni?
Gemini Omni is Google’s model for video generation and editing. You describe what you want in words, or you give it references — a product photo, a character image, a short motion clip, an audio cue — and it produces a video. Unlike a text-to-video model that only works from a prompt, Omni is built around references: it is designed to keep a specific product, person or brand looking the same from the first frame to the last.
The “Flash” in the name signals speed and cost. It is the fast, cheaper branch of Google’s video models, aimed at the everyday videos businesses make in volume: product videos, virtual try-ons, social ads, explainers. The slower, more cinematic Veo models sit above it for work where quality matters more than throughput. Omni 1.2 Flash is the latest version of that fast branch, and the first to be generally available on Google Cloud.
General availability matters because it changes who can rely on the model. A preview or experimental release can change or disappear; a generally available service comes with Google Cloud’s enterprise infrastructure, security and support, plus Provisioned Throughput for businesses that need guaranteed capacity. That is why Google’s announcement was pitched squarely at businesses and developers, not just at hobbyists trying out a new toy.
What changed in Gemini Omni 1.2 Flash?
Four things changed against Omni 1.1, and they are worth taking one at a time, because each one answers a different frustration with the older model.
1. Clips are 50% longer
The maximum clip length rose from 10 seconds to 15 seconds — a 50% increase. You can also set the duration in one-second steps, anywhere from 3 to 15 seconds, so a clip can be timed to fit a specific placement instead of being cut down afterwards. Fifteen seconds is the length of a standard social video ad, and it is long enough for a complete product explainer or a full scene in a single generation, rather than two clips stitched together.
2. Editing happens in conversation
Previously, fixing a detail usually meant generating the whole clip again. Omni 1.2 Flash supports multi-turn conversational editing: you keep the same clip and ask for changes in plain language — swap the product, change the background, adjust the lighting or weather, switch the aspect ratio — and the model applies the edit while keeping everything else intact. The editing works across full 15-second clips, so a marketing team can produce variants of one ad for different audiences or placements without re-briefing the model each time. Google’s own example walks one scene through five background changes — a striped pavilion tent, an oak forest, a stable with a white horse, a market fair, a stone castle — while the characters’ performances and camera framing stay put.
3. References are followed more closely
You can guide generation with character, product or style images, motion clips of up to three seconds, and audio cues, and the new model holds onto those details more tightly through the whole clip. Faces, garments, packaging typography and brand colours stay consistent across camera angles. In practice, that is what makes virtual try-on clips work: Google describes combining one model photo with front-and-back shots of three outfits to produce a 15-second try-on sequence, with the model’s identity and each garment’s cut, drape and back details staying consistent through 360-degree turns and match cuts. Audio references are gated behind allowlisting, so access to that feature is limited for now.
Consistency also improved between frames. Google says the model shows less visual drift or morphing across motion and camera transitions, and holds products, characters, props and environments steady whether you are interpolating between a first and last keyframe or extending a clip into a longer sequence.
4. The price fell 30%
The headline price is $0.07 per second for 720p video, down 30% from Omni 1.1. By simple arithmetic, the older model worked out at about $0.10 a second. Google says similar savings apply at other resolutions, though it has only published the 720p figure. The workflow also keeps costs down in a second way: you can draft at 360p or 720p, fix details with a follow-up edit instead of re-generating, and only upscale your best creations to 1080p or 4K.
Gemini Omni 1.2 Flash vs Omni 1.1 at a glance
| Omni 1.1 Flash | Omni 1.2 Flash | |
|---|---|---|
| Maximum clip length | 10 seconds | 15 seconds |
| Duration control | Fixed lengths | 3–15 seconds in 1-second steps |
| Price (720p) | About $0.10 per second (implied by the 30% cut) | $0.07 per second |
| Editing | Regenerate to change | Multi-turn conversational editing |
| Reference following | Standard | Enhanced — images, video and audio held more closely |
| Availability | Preview | Generally available on the Gemini API, Google Cloud and Gemini for business |
| Speech | Basic dialogue | More natural dialogue, clearer lip-sync, distinct multi-speaker voices |
| Watermark | SynthID | SynthID (imperceptible) |
How to use Gemini Omni 1.2 Flash
There are two routes in: the developer route and the app route. Which one fits depends on whether you are building a product or making a video.
For developers: the Gemini API and Google Cloud
Developers can prototype in Google AI Studio, which offers starter apps and the developer documentation, then build production systems through the Gemini API or Google Cloud‘s Vertex AI platform. The typical workflow Google describes is: draft fast at 360p or 720p, refine with conversational edits, then upscale the best results to 1080p or 4K. Because the model is generally available, businesses can also buy Provisioned Throughput on Google Cloud for predictable capacity at scale.
Creative platforms are already embedding the model directly, so many users will meet it inside tools they already use — Artlist, Krea, Picsart, Photoroom and Runway were all named in Google’s announcement, and Photoroom said it uses Omni to power its Video Ad Maker and Product Animator for sellers.
For everyone else: Google Vids, Flow, the Gemini app and Google Ads
If you do not write code, the model is rolling out across Google’s own products. In Google Vids, teams can generate 15-second clips at no additional cost with a Google account or a Google Workspace subscription, then combine them with avatars, music and voiceover into product explainers, teasers, social posts and sales pitches. In Flow, you direct scenes and apply multi-turn edits. In Google Ads, you can create 15-second videos to scale campaigns. In the Gemini app, 15-second video creation is rolling out gradually, reaching Google AI Ultra subscribers first.
A practical starting point for a first try is Google Vids: pick a product photo you already have, ask for a short clip, and see how well the model holds the product’s shape, colours and packaging text. That one test tells you more about whether it fits your use case than any demo video.
What does Gemini Omni 1.2 Flash cost?
The only published price is $0.07 per second for 720p video. Worked examples at that rate:
- A 5-second clip at 720p: 5 × $0.07 = $0.35
- A 10-second clip at 720p: 10 × $0.07 = $0.70
- A 15-second clip at 720p: 15 × $0.07 = $1.05
- One hundred 15-second clips at 720p: 100 × $1.05 = $105
Two things are worth noting about those numbers. First, the price is per second of output, so the cost of a failed or discarded clip is the same as a good one — drafting at lower resolution and editing in conversation rather than re-generating is how you keep the bill down. Second, the per-second rate means longer clips cost proportionally more: a 15-second clip costs 50% more than a 10-second one at the same resolution, even though the rate itself fell. Google has said similar savings apply at other resolutions but has not published those prices, so do not budget on a specific 1080p or 4K rate until it is confirmed.
In Google Vids, Flow and the Gemini app, 15-second clips are included at no additional cost to Google account and Workspace users — the per-second pricing applies to the API and Google Cloud route, not to casual use inside Google’s apps.
Who is Gemini Omni 1.2 Flash for?
Google’s own framing is clear: businesses, developers and creators producing videos at scale. The strongest fits are the ones where you already have the raw material — product photos, a brand style, a character sheet — and need many short videos from it. That includes online retailers turning catalogue photos into product spins and try-on clips, marketers producing ad variants for different placements, and small businesses that cannot afford a video shoot for every product.
It is also useful for anyone who needs consistency across a series: a character who must look the same in ten clips, a product whose packaging text must stay legible, a brand colour that must not drift. That is the job the improved reference following is built for.
It is less obviously the right tool for a single cinematic short film, where a slower, higher-quality model may matter more than speed and price, or for fast sports action and complex physics, which Google itself lists as works in progress.
How does it compare with rivals?
The most direct comparison is on clip length. ByteDance’s Seedance 2.5 generates up to 30 seconds in a single pass — twice Omni 1.2 Flash’s 15 — according to Krea’s tutorial for the model. Omni’s answer is not length but workflow: conversational editing, tighter reference following, and a lower per-second price at 720p. Seedance’s 720p rate on BytePlus ModelArk has been reported at about $0.23 per second, against Omni’s $0.07 — though prices change and should be checked at the source before budgeting.
The honest comparison is that no single number settles it. A model that makes a 30-second clip is better for a long single take; a model that edits in conversation and holds a product’s packaging steady across angles is better for a catalogue of ads. The right choice depends on whether your bottleneck is length, consistency, or cost per clip.
What are the limits?
Google is unusually candid about what does not work yet. Fast sports action, rapid script changes and complex physics remain works in progress. The company published no quantitative benchmarks for the new model — only side-by-side examples showing it following references more closely than Omni 1.1. That means there is no independent score to compare against rivals; the examples are chosen by Google, and your own test matters more than theirs.
Other limits to keep in mind: audio references are gated behind allowlisting, so many developers cannot use them yet; the Gemini app rollout is gradual and reaches Google AI Ultra subscribers first; and every output carries an imperceptible SynthID watermark, which is a feature for transparency but means the video is marked as AI-generated.
Finally, the 15-second ceiling is still a ceiling. A full 30-second or 60-second ad still needs multiple clips joined together, and holding consistency across those joins is a separate problem from holding it inside one clip.
Frequently asked questions
Is Gemini Omni 1.2 Flash free?
Not through the API: it costs $0.07 per second for 720p video. But 15-second clips are included at no additional cost in Google Vids for Google account and Workspace users, and the model is rolling out in Flow, Google Ads and the Gemini app.
How long can Gemini Omni 1.2 Flash videos be?
Up to 15 seconds, in one-second steps from 3 to 15 seconds. The previous version, Omni 1.1, capped clips at 10 seconds.
Can I edit a video after generating it?
Yes. Multi-turn conversational editing lets you refine the same clip with follow-up prompts — changing the product, background, lighting, weather or aspect ratio — instead of generating it again from scratch.
Does it add a watermark?
Yes. Outputs carry an imperceptible SynthID watermark, Google’s marker for AI-generated content.
Where can I try it?
Developers can prototype in Google AI Studio and build with the Gemini API or on Google Cloud. Everyone else can use it in Google Vids, Flow and Google Ads, with a gradual rollout in the Gemini app for Google AI Ultra subscribers first.
For the day’s news on this release, see our report on informopedia.net.
More in AI: What Is Tencent Hy4? · How to Make GIFs in ChatGPT Images 2.6
