Some links below are affiliate links. If you buy through them we may earn a commission at no extra cost to you. We only list tools we would use ourselves. How we make money.
Gemini Omni 1.2 Flash is Google's newest AI video model, released on 5 November 2026. One prompt now gets you a clip of up to 15 seconds with sound, up from 10 seconds with Omni 1.1, and Google cut the price by 30%. Its pricing page lists about $0.07 per second of 720p video in the Gemini API, roughly $1.04 for a full 15-second clip. It makes vertical 9:16 or landscape 16:9 video, generates dialogue, music and sound effects in the same pass, and lets you fix a clip by asking for changes instead of starting over.
You can use it in the Gemini API and Google AI Studio (paid tier), the Google Cloud console, Flow, Google Vids, Google Ads and the Gemini app, where it is rolling out to Google AI Ultra subscribers. Below: what changed, prices, first-clip steps, eight prompts for Shorts and Reels, and a comparison with Veo, Kling, Seedance and Sora. Honest note: we haven't generated clips with it yet; this is research from Google's pages and launch-day reporting, read on 6 November 2026.
What's new in Gemini Omni 1.2 Flash?
Here's what Google and The Decoder say changed from Gemini Omni 1.1 Flash:
- 15-second clips. Any whole second from 3 to 15, up from a 10-second maximum. The API still defaults to 10 seconds unless you set the duration.
- 30% cheaper. About $0.07 per second of 720p video, against about $0.10 for Omni 1.1, according to Google's pricing page.
- Edit by conversation across the whole clip. Swap a product or background, change the lighting or weather, or switch the aspect ratio in a follow-up, instead of regenerating everything.
- Better reference following. Guide it with character, product or style images and motion clips of up to 3 seconds. The API takes up to 6 reference images (3 per character, for up to 2 characters). The Decoder sums up the promise as consistent faces, clothing, packaging and brand colours.
- Draft cheap, finish sharp. Draft at 360p or 720p, then make the keeper at 1080p or 4K (both upscaled).
- Better speech and camera work. Google promises more natural dialogue, clearer lip-sync, distinct voices for several speakers and "significantly fewer unwanted jump cuts".
- Longer sequences in the API. Extend a generated clip by 3 to 10 seconds per turn, up to 30 seconds in total, or fill the motion between a first and last keyframe (3 to 10 seconds).
Google's DeepMind page shows raters preferring 1.2 over 1.1 in every category it charts (62% to 38% on text-to-video visual consistency), without saying who rated or how many clips. Treat that as a vendor claim. Every output carries Google's invisible SynthID watermark.
Where can you use Gemini Omni video, and who gets access?
| Where | Who it's for | What Google says |
|---|---|---|
| Gemini app | Google AI Ultra subscribers | 15-second videos, "rolling out" to Ultra subscribers |
| Google Vids | Google account or Workspace users | 15-second clips "at no additional cost" |
| Flow | Creators building scenes and short films | Direct scenes and multi-turn edits, including dialogue between several characters |
| Gemini API and Google AI Studio | Developers and power users | Generally available on the paid tier; model ID gemini-omni-1.2-flash; free tier "Not available" |
| Google Cloud console | Businesses | Generally available, with Provisioned Throughput |
| Google Ads | Advertisers | Use it to "scale video campaigns" |
The fine print: "Features vary by tier and geography," says Google. In the EEA, Switzerland and the UK you can't edit or extend uploaded videos (generated clips are fine), images of certain recognisable people can't be uploaded, and only English is fully supported. YouTube isn't mentioned; for the free AI tools inside the Shorts app, see our best AI video generator for YouTube Shorts comparison.
Gemini Omni price: what does a 15-second AI video cost?
The Gemini API bills by tokens: $12.00 per million video output tokens (Omni 1.1 was $17.50), plus $1.50 per million input tokens for your text, images and video. Google's pricing page converts that into per-second rates:
| Resolution | Per second (Google's figure) | One 15-second clip |
|---|---|---|
| 360p (drafts) | $0.023 | about $0.35 (our maths) |
| 720p (default) | about $0.07 | about $1.04 (Google's figure) |
| 1080p, upscaled | $0.104 | about $1.56 (our maths) |
| 4K, upscaled | $0.209 | about $3.14 (our maths) |
- Output only. Input tokens come on top, and Google gives no per-second figure for them.
- No free API tier. The free-tier cells read "Not available", so expect to add billing.
- Omni 1.1 is being retired. From 3 December 2026, API calls to
gemini-omni-1.1-flashgo to 1.2 and are billed at 1.2 rates. - Apps are priced differently. Google Vids clips come "at no additional cost" with a Google account or Workspace subscription. The Gemini app needs Google AI Ultra; check the current price on Google's plans page.
A realistic budget per finished clip: five 15-second drafts at 360p plus one fresh 1080p render (the docs only describe setting resolution at generation time) comes to about $3.29 before input tokens. Prices change, so check the current price on Google's pricing page before quoting a client.
How to make your first 15-second AI video with Gemini Omni
Option A: the Gemini app or Google Vids
With Google AI Ultra, describe your video in the Gemini app and say the length (up to 15 seconds) and that it should be vertical. With a regular Google account, Google says Vids makes 15-second clips at no additional cost, the cheapest way to try the model. No 15-second option yet? The rollout hasn't reached your account.
Option B: Google AI Studio or the API, paying per second
- Set up a paid-tier API key in Google AI Studio and choose
gemini-omni-1.2-flash. - Set the format. In
response_format, setdurationto"15s"andresolutionto"360p"for drafts. Setaspect_ratioto"9:16"for Shorts, Reels and TikTok (16:9 is the default). The duration you set overrides any length in your prompt. - Write the prompt in time blocks. Google's docs support timecodes like
[0-5s] ... [5-10s] ... [10-15s] ..., the easiest way to pace 15 seconds. Cover scene, camera movement, lighting and mood, plus a sound line. - Ask for one shot if you want one. Google says a single scene must be prompted, for example "In a single continuous shot. No scene cuts."
- Fix with short follow-ups. Point a new turn at the previous one (
previous_interaction_idin the API) and describe one change. Google warns that "overly descriptive prompts can lead to unintended changes" when editing. - Render the keeper at 720p, 1080p or 4K, add captions and hook text in your editor, and label it before posting.
Shape a prompt before spending credits in our free prompt studio.
8 copy-paste Gemini Omni prompts for Shorts and Reels
Each is written for a vertical 9:16 clip with the duration set to 15 seconds. Swap in your niche, and keep real people and other people's brands out.
1. Scroll-stopping hook
[0-3s] Extreme close-up of a coffee mug on a kitchen table, soft morning light. [3-8s] The camera pulls back fast to reveal the table floating above the clouds at sunrise. [8-15s] Slow orbit as steam curls from the mug, golden light. Single continuous shot, no scene cuts. Sound: room tone, a deep whoosh at 3 seconds, then wind and soft ambient music. No dialogue.
2. Product reveal (add your product photo as a reference)
Keep the reference product's shape, label and colours exactly. [0-4s] Dark studio, a beam of light sweeps across an empty black pedestal. [4-10s] The product rises into the light and turns slowly, water droplets catching highlights. [10-15s] Push in to a close-up of the label, then hold. Sound: low synth pulse, a crisp chime when the light hits the product. No dialogue.
3. Seasonal holiday clip
A gingerbread house on a frosted windowsill at night, snow falling outside, blurred warm string lights. [0-5s] Slow push-in. [5-10s] A tiny light switches on inside the house. [10-15s] Snow thickens as the camera settles on the glowing window. Single continuous shot. Sound: crackling fireplace, gentle music-box melody.
4. Talking-object ad
A worn running shoe on a park bench at sunrise, with a friendly cartoon face formed by its laces. [0-6s] It looks at the camera and says in a cheerful, tired voice: "Three hundred kilometres together, and you still haven't cleaned me." [6-11s] It sighs and glances at a shiny new pair beside it. [11-15s] Close-up as it winks. Sound: birdsong, a soft comedic sting on the wink.
5. Travel b-roll
FPV drone flight at golden hour. [0-5s] Gliding low over turquoise water toward a whitewashed coastal village. [5-10s] Rising smoothly over terracotta rooftops. [10-15s] Revealing a wide bay with sailboats and mountains. One continuous flight, no cuts. Sound: waves, seagulls, rushing wind. No music, no dialogue.
6. Cinematic loop
A paper boat drifting along a neon-lit canal in a rainy city at night, electric blue and warm orange reflections, raindrops rippling the water. Slow tracking shot at water level following the boat. The final frame matches the opening composition so the clip can loop. No scene cuts. Sound: steady rain, distant city hum, low synth pad.
A prompt alone won't always loop cleanly. In the API, first-and-last-frame mode with the same image at both ends is the cleaner trick, capped at 10 seconds.
7. Before-and-after transformation
Fixed wide shot of a cluttered, dim studio apartment. [0-5s] Time-lapse: boxes and clothes clear away in quick bursts. [5-10s] Walls turn warm white, plants and a wooden shelf appear, sunlight floods in. [10-15s] Slow push-in on a cosy reading chair in the finished room. No scene cuts. Sound: quick whooshes, then calm acoustic guitar.
8. Satisfying miniature world
Macro shot of a thumb-sized chef in a miniature wooden kitchen. [0-5s] Batter pours onto a coin-sized pan in slow motion. [5-10s] The chef flips the pancake high into the air. [10-15s] It lands on a tiny stack as a drop of syrup rolls down. Shallow depth of field, warm light. Single continuous shot. Sound: sizzling, a flip whoosh, a tiny cheerful "ta-da!".
Gemini Omni 1.2 Flash vs Veo, Kling, Seedance and Sora
Here's how Google's new model compares on short-form specs, from each maker's own pages. Prices are left out because each service bills differently.
| Model | Maker | Longest clip per generation | Native audio | Vertical 9:16 | Where creators use it |
|---|---|---|---|---|---|
| Gemini Omni 1.2 Flash | 15 s (API extends to 30 s total) | Yes | Yes | Gemini app (AI Ultra), Vids, Flow, Gemini API, Google Cloud | |
| Veo 3.1 | 8 s (Extend chains clips) | Yes | Yes | Flow, Vertex AI; listed on Higgsfield and getimg.ai | |
| Kling 4.0 | Kuaishou | 30 s | Yes, stereo | Yes | Kling AI; listed on Higgsfield and getimg.ai |
| Seedance 2.5 | ByteDance | 30 s | Yes | Yes | Dreamina, CapCut (subscribers in some regions), BytePlus API; listed on Higgsfield and getimg.ai |
| Sora 2 | OpenAI | Discontinued | n/a | n/a | App closed 26 April 2026, API 24 September 2026 |
Kling 4.0 and Seedance 2.5 already make 30-second clips, so 15 seconds is a step up for Google, not a market record. Google is also consolidating: its AI Studio docs say the Veo 3.1 preview in the Gemini API has been shut down and point developers to Gemini Omni Flash. Omni's edge, per Google, is editing by conversation and a low draft price.
To see how one brief comes out in different models before committing, use a multi-model platform. When we checked on 6 November 2026, Higgsfield listed Kling 4.0, Seedance 2.5 and Veo 3.1 side by side, plus Gemini Omni Flash 1.1, the previous version, not 1.2. Our link gets you up to 50% off for 3 hours after you arrive.
getimg.ai does the same on one credit balance: its video page lists Kling 4.0, Seedance 2.5, Google Veo 3.1 and Gemini Omni Flash 1.1 (not 1.2 yet), all with native audio. For multi-model studios versus paying a model maker directly, read Higgsfield vs Kling: which one to pay for.
Is a 15-second AI video enough? Honest limits
What to expect, honestly:
- 15 seconds is one beat, not a story. A 45 to 60-second Short still means stitching three or four clips, and every cut is another chance for a face or product to drift.
- Consistency across shots varies. References and conversational edits help, but Google says it is still improving "fast sports action, rapid script changes and complex physics".
- Costs add up when you iterate. A 15-second 720p clip costs about a dollar in the API, and you'll rarely keep the first one. Draft at 360p and set a budget per finished clip.
- Label AI content where platforms ask. YouTube requires you to disclose realistic AI-generated or altered content in YouTube Studio. TikTok requires a label on AI-generated content showing realistic-looking scenes or people. Meta requires its AI disclosure tool for photorealistic video and may apply penalties if you skip it. The invisible SynthID watermark doesn't replace these labels.
- Not tested by us yet. This is research from Google's pages and launch reporting; your results depend on your prompts and subject.
Bottom line: the real gain for short-form creators is fewer regenerations. Start in Google Vids or with a few 360p drafts, keep the prompts that work, and pay for 1080p only when a clip has earned it.
FAQ
What is Gemini Omni 1.2 Flash?
Google's video generation and editing model, released on 5 November 2026. It makes 3 to 15-second clips with native audio in 9:16 or 16:9, and you can change a finished clip by describing the edit instead of regenerating it.
How long can a Gemini Omni video be?
Up to 15 seconds per generation, set in one-second steps from 3 seconds. In the Gemini API you can extend a generated clip by 3 to 10 seconds per turn, up to 30 seconds in total.
How much does Gemini Omni 1.2 Flash cost?
In the Gemini API, Google lists about $0.07 per second at 720p (about $1.04 for 15 seconds), $0.023 at 360p, $0.104 at 1080p and $0.209 at 4K. That is 30% less than Omni 1.1. Check the current price on Google's pricing page.
Is Gemini Omni 1.2 Flash free?
Not in the API, where the free tier lists it as "Not available". Google says Google Vids makes 15-second clips at no additional cost with a Google account or Workspace subscription, and the Gemini app version is rolling out to Google AI Ultra subscribers.
Can I use Gemini Omni 1.2 Flash on Higgsfield or getimg.ai?
Not as of 6 November 2026. Both list the previous Gemini Omni Flash 1.1 next to Kling 4.0, Seedance 2.5 and Veo 3.1. Check their model menus for updates.
Does Gemini Omni make vertical videos for Shorts and Reels?
Yes. Google's API docs list 9:16 portrait and 16:9 landscape, with landscape as the default, so set 9:16 for Shorts, Reels and TikTok.
Sources
- Google: Gemini Omni 1.2 Flash, fast and cost-effective with longer clips (5 Nov 2026)
- Google AI Studio: Gemini API pricing (Gemini Omni 1.2 Flash)
- Google AI Studio: Gemini Omni video generation docs
- Google Cloud: Gemini Omni 1.2 Flash is generally available
- Google DeepMind: Gemini Omni model page
- The Decoder: Gemini Omni 1.2 Flash stretches AI video clips to 15 seconds and cuts pricing by 30 percent
- Google AI Studio: video generation docs (Veo 3.1 preview shutdown)
- Google Cloud: Veo 3.1 model reference
- Google: Veo updates in Flow (Extend)
- Kling AI: Kling 4.0 model user guide
- Kling AI: Kling 4.0 vs 3.0
- ByteDance Seed: Introducing Seedance 2.5
- BytePlus: Seedance 2.5 API docs
- CapCut newsroom: Dreamina Seedance 2.5 rollout
- OpenAI: Sora 2 page (product discontinued 26 April 2026)
- OpenAI Help Center: What to know about the Sora discontinuation
- Higgsfield: Gemini Omni Flash page
- Higgsfield: AI video generator models
- getimg.ai: AI video generator
- getimg.ai: Gemini Omni Flash 1.1 video generator
- YouTube Help: Disclosing use of GenAI content
- TikTok Community Guidelines: Integrity and authenticity
- Meta: How Meta is preparing for the 2026 US midterms (AI info labels)