How to Use Gemini Omni 1.2 Flash for 15-Second AI Videos

Google made Gemini Omni 1.2 Flash generally available on 5 November 2026, and it is the first version of its video model that feels built for finished work rather than demos. Clips now run to 15 seconds, you can edit the same clip in plain words instead of starting again, and the price fell 30% to about $0.07 a second at 720p. This guide walks through generating your first clip, editing it, and working out what a finished video really costs — with one honest caveat up front: I have not tested the model hands-on, so every fact below comes from Google’s own announcement and pricing pages, checked on 7 November 2026.

Quick answer: Gemini Omni 1.2 Flash makes 15-second AI videos with sound for $0.07 a second at 720p. Here's how to generate, edit and extend a clip — and what it can't do yet.

What is Gemini Omni 1.2 Flash?

Gemini Omni 1.2 Flash is Google’s generally available AI video model — model ID gemini-omni-1.2-flash — that generates and edits video with native audio.

Google announced it on 5 November 2026 through a DeepMind blog post by product manager Anish Nangia, with general availability across the Gemini API and Gemini for business the same day. It takes text, images and video as input and returns video with sound, at 360p, 720p, 1080p or 4K, in 16:9 or 9:16, at 24 frames per second. It is a video model, not a text model — do not confuse it with the Gemini 4 family, which covers Google’s text models. Our Gemini 4 Carbon explainer and the Gemini 4 Argon vs Carbon comparison cover that side of the family.

What changed from Omni 1.1 Flash:

Omni 1.1 Flash Omni 1.2 Flash
Longest single clip 10 seconds 15 seconds
Video output price $17.50 per 1M tokens $12.00 per 1M tokens
Price at 720p About $0.10 a second (implied) About $0.07 a second
Editing Start over for changes Multi-turn conversational editing, first-and-last-frame interpolation, clip extension
Reference images Fewer, weaker consistency Up to 6, with better character consistency (Google’s claim)
Availability Preview Generally available, Gemini API + Gemini for business

Two numbers deserve a second look. First, the 15-second cap: ten seconds is a teaser, but fifteen seconds is a complete product explainer, a full social ad spot, or a Short with a beginning, a middle and an end. Second, the price: Google announced a 30% cut to $0.07 a second, which implies Omni 1.1 sat at about $0.10 — that old figure is arithmetic, not a quote, because Google did not print it in the announcement.

How to generate your first clip

Pick your route first, because the price and the limits change with it — there are three ways in.

  1. Gemini app (easiest). Open the Gemini app and describe the clip you want. Fifteen-second videos are rolling out to Google AI Ultra subscribers, so if you are not on Ultra — or the rollout has not reached you yet — the app will not show the longer duration. This is the route for a first test, not for bulk work.
  2. Google Vids (cheapest). Google Vids generates 15-second clips at no additional cost with a Google account or Workspace subscription, and Google says the full rollout should take one to three days from 5 November for the feature to become visible. If you already live in Workspace, start here.
  3. Gemini API (most control). On the paid tier of the Gemini API, call model ID gemini-omni-1.2-flash through the Interactions API with an API key from Google AI Studio. There is no free tier for this model on the API — that matters for budgeting, and it is the only route where every second is metered.

Once you have picked a route, the workflow is the same:

  1. Write one scene, not a film. A 15-second clip holds a single moment. Put the subject, the action, the setting, the camera move, the light and the sound in that order — for example: “A barista pours latte art in a sunlit café, slow dolly-in, warm morning light, soft café murmur.”
  2. Set the duration. Clips run 3 to 15 seconds in one-second steps, and default to 10. Ask for 15 only when the scene needs it; a five-second clip that lands is worth more than fifteen seconds of drift.
  3. Pick the shape. Use 16:9 for YouTube and embeds, 9:16 for Shorts, Reels and TikTok. Resolution is a cost decision as much as a quality one — see the pricing section below.
  4. Generate, watch, judge. Check three things before anything else: does the subject stay the same from the first frame to the last, does the motion look deliberate rather than smeared, and does the audio match the picture? If any of those fail, change the prompt — do not re-roll the same words hoping for luck.

Three copy-paste prompts to start from — adapt the bracketed parts, keep the structure:

How to edit a clip without starting over

Multi-turn editing is the real upgrade in 1.2 — you describe the change in plain words and the model revises the same scene.

What does a clip actually cost?

At 720p, a 15-second clip costs about $1.04 in video-output tokens — but that is the cost of one attempt, not of a finished video.

Google bills video output by the token: $12 per million tokens, down from $17.50 for Omni 1.1. At 720p, Google counts 5,792 tokens per second, which is how the $0.07-a-second shorthand is derived. Input tokens cost $1.50 per million and text output $9 per million, so the all-in bill is slightly higher than the video-output figure alone.

Resolution Per second 5-second clip 15-second clip
360p $0.023 $0.12 $0.35
720p About $0.07 $0.35 About $1.04
1080p (upsampled) About $0.104 $0.52 $1.56
4K (upsampled) About $0.209 $1.05 $3.14

Those figures come from Google’s pricing page, via independent explainers that checked it on 6 November 2026. The 15-second 720p figure works out at $1.04 from the token maths (86,880 tokens at $12 per million); multiplying the rounded $0.07 rate by 15 gives $1.05. Either way, a finished clip costs about a dollar at 720p.

Here is the part most first-time users miss, and it is the original point of this guide: every attempt bills, so a finished clip costs drafts times seconds, not seconds alone. If you generate eight versions before one is usable, that is eight times the table price. The fix is to draft cheap and render dear: test prompts at 360p, where a 15-second draft costs about $0.35, and only render the keeper at 720p or 4K. Nine 15-second drafts at 360p cost about the same as one 15-second render at 4K ($3.15 versus $3.14). Rule of thumb: never draft at the resolution you plan to ship. Google also doubled the paid-tier rate limits for this model, so bulk drafting is less likely to hit a cap — but the meter still runs.

Where it fits — and where it falls short

Fifteen seconds plus real editing makes this a tool for short-form work, not a replacement for a camera — and the limits are specific enough to plan around.

Route Pick it if… Watch out for…
Gemini app You want the fastest first test 15 seconds is an Ultra-subscriber rollout, not live for everyone
Google Vids You want 15-second clips without a per-second bill It is a Workspace/personal-account feature, not an API
Gemini API You need control, volume or automation Paid tier only — no free tier — and every attempt bills

The limits, all from Google’s own documentation:

Our take

The price cut is the headline, but the editing is the story. A 30% cheaper second matters less than the fact that a bad first draft is no longer a dead end — multi-turn edits, frame interpolation and extension turn generation from a slot machine into a tool you can steer. Fifteen seconds is also a genuine threshold: it is the first length where a complete ad, a product demo or a Short can exist in one clip instead of three stitched together.

The honest counterweight is that nothing here is independently verified yet. No benchmarks, no free tier on the API, and a rollout that reaches Ultra subscribers first means most readers will be testing, not shipping, for a while. Our advice: draft at 360p in Vids or the app, keep the prompt to one scene, and only pay for 720p or 4K once the clip already works. If the clip does not work at 360p, a more expensive render will not fix it.

FAQ

Is Gemini Omni 1.2 Flash free?

Not on the Gemini API — there is no free tier for this model, and it sits on the paid tier only. The free routes are Google Vids, where 15-second clips come at no additional cost with a Google account or Workspace subscription, and the Gemini app, where 15-second videos are rolling out to Google AI Ultra subscribers.

How long can a Gemini Omni 1.2 Flash video be?

Between 3 and 15 seconds in a single generation, in one-second steps, with 10 seconds as the default. You can extend a clip by 3 to 10 seconds per turn up to 30 seconds in total, and first-and-last-frame interpolation works for clips of 3 to 10 seconds.

How much does a 15-second video cost?

About $1.04 in video-output tokens at 720p (86,880 tokens at $12 per million), or about $1.05 if you multiply the rounded $0.07-a-second rate by 15. Input tokens ($1.50 per million) and text output ($9 per million) add a little on top, and every draft bills — so budget for attempts, not just the final clip.

Can it keep the same character across clips?

Google says yes — up to 6 reference images with improved character consistency. But Google published no benchmark for this, so treat it as a claim to test with your own reference set before building a series around a recurring character.

What happens to Omni 1.1 on 3 December 2026?

From that date, requests to gemini-omni-1.1-flash automatically route to gemini-omni-1.2-flash and are billed at the newer model’s rates, according to Google’s pricing page. If your code still calls the old model ID, it will keep working — at 1.2 prices.

Sources and methodology

Method: desk research from Google’s announcement and pricing pages plus two independent explainers that checked those pages on 6 November 2026; all claims verified 7 November 2026. No hands-on testing was performed — the model was not available to test from India at the time of writing, and quality or consistency claims are labelled as Google’s throughout. Prices are Google’s published Standard token rates and cover video output only unless stated otherwise.

Leave a Reply

Your email address will not be published. Required fields are marked *