Gemini Omni Flash: Google’s Talk-to-Edit AI Video Model Explained

Gemini Omni 1.2 Flash is Google’s talk-to-edit AI video model, and since November 5, 2026 it costs $0.07 per second of 720p video through the Gemini API. That is about $0.70 for a 10-second clip and $1.04 for a full 15-second clip, 30% less than Omni 1.1 Flash. You describe the change you want in plain language and the model rewrites the shot, so there is no timeline to drag clips around on.

This guide covers what changed in Omni 1.2 Flash and what it costs at every resolution. It also shows where you can use it today: the Gemini app, Google Flow, Google Vids, YouTube Shorts, and the API. You will also see where its real limits sit and how it pairs with Google’s fast image model. A comparison with Veo closes the guide, so you can decide whether Omni Flash belongs in your creative workflow.

The Key Takeaways

• $0.07 per second of 720p video through the Gemini API, about $1.04 for a 15-second clip, down 30% from Omni 1.1 Flash.
• Released on November 5, 2026 and generally available on the Gemini API and Google Cloud. The first Omni Flash was announced at Google I/O 2026 on May 19.
• Conversational editing means you refine a clip by describing changes, not by editing a timeline, and in 1.2 it works across the full 15 seconds.
• Clips run 3 to 15 seconds in 16:9 or 9:16, with native audio and a SynthID watermark on every generation. Scene extension stretches a video to 30 seconds in total.
• Free routes exist: three 15-second videos a month in Google Vids, 50 daily credits in Google Flow, and YouTube Shorts for users 18 and over. The Gemini app needs a Google AI plan, and 15-second clips there are AI Ultra only.

What’s New in Gemini Omni Flash

Gemini Omni 1.2 Flash makes clips longer and cheaper. Google raised the native clip length from 10 to 15 seconds and cut the API video price by 30%, according to its Omni 1.2 Flash launch post. You can now set the duration in one-second steps anywhere from 3 to 15 seconds.

Omni 1.2 Flash also holds a scene together better. Google says characters, products and backgrounds drift and morph less across camera moves, and that complex camera paths like FPV flythroughs and orbits show fewer unwanted jump cuts. Faces, garments, packaging text and brand colors stay closer to the reference images you upload. Dialogue got an upgrade too, with more natural speech, clearer lip-sync and separate voices for several speakers.

Google is open about what Omni 1.2 Flash still gets wrong. Its launch post names fast sports action, rapid script changes and complex physics as areas still in development. Developers on Omni 1.1 Flash do not need to migrate by hand. The Gemini API changelog says requests to gemini-omni-1.1-flash route to 1.2 automatically on December 3, 2026, billed at the 1.2 rate.

What Is Gemini Omni Flash?

Gemini Omni Flash is Google’s any-to-any AI video model. It takes text, images, and video as input and generates or edits short clips through natural conversation. You describe a change in plain language and the model applies it while keeping the rest of the shot intact, so a rough idea becomes a finished clip in a handful of turns. The current version is Omni 1.2 Flash, released on November 5, 2026.

The “any-to-any” label matters. Older tools did one job, like text to video or image to video, and stopped there. Omni Flash mixes those inputs in a single request, so you can hand it a reference photo, add a written prompt, and steer the result with follow-up instructions. It is the fast, cheaper tier of Google’s wider Gemini Omni family, built for speed and volume rather than maximum cinematic polish. For the chat side of the family, our Gemini 3.5 review and Pro release tracker covers Flash and the still-unreleased Pro.

Under the hood it runs on Google’s new Interactions API, which replaces the older single-request pattern with a back-and-forth session. That session is what makes talk-to-edit possible, because the model remembers the clip you are working on and layers each new instruction on top. Omni Flash left preview with version 1.1 on August 27, 2026, and Omni 1.2 Flash is generally available on the Gemini API and Google Cloud.

How Talk-to-Edit Video Works

The headline feature is conversational editing. You generate a clip, watch it, then type or say what should change, and the model returns a revised version. There is no timeline, no keyframes, and no layer panel, which is a different way to work for anyone who has wrestled with traditional video software.

Say you generate a 6-second shot of a coffee cup on a desk. You can follow up with “make it morning light,” then “add steam rising from the cup,” then “pan slowly to the left,” and each instruction updates the same clip. The model tries to preserve the parts you did not mention, so the cup stays where it was while the lighting and motion change around it. In Omni 1.2 Flash those edits work across a full 15-second clip. You can swap a product, a background or the weather in a finished ad without generating it again.

Every clip ships with native audio generated from your description and an invisible SynthID watermark that flags it as AI-made. That watermark is not optional, which is worth knowing if you plan to publish the output commercially. You can read Google’s own breakdown of the first version in its Gemini Omni Flash announcement.

Gemini Omni Flash Specs and Pricing

Gemini Omni 1.2 Flash generates 720p video by default at $0.07 per second, in clips of 3 to 15 seconds. It trades native resolution for speed and a low price, which is the whole point of a “Flash” tier. Here is what you actually get.

Pricing and free access

Gemini Omni 1.2 Flash costs $0.07 per second of 720p video on the paid tier of the Gemini API. Google bills it as $12.00 per million output tokens at 5,792 tokens per second, so a 10-second clip costs about $0.70 and a 15-second clip about $1.04, before any regeneration. The Gemini API pricing page lists the other resolutions too.

The 360p draft mode is the cheapest way to iterate. Google added it with Omni 1.1 Flash on 27 August 2026, and at $0.023 a second it costs about a third of the 720p rate. Black Forest Labs prices FLUX 3’s draft mode at roughly the same ratio: $0.06 a second against $0.17 for HD. The same 1.1 release added scene extension and upscaling to 1080p or 4K, so the 720p ceiling is a generation limit rather than a delivery one. In Omni 1.2 Flash, extension reads the last 10 seconds of a video and adds 3 to 10 seconds per turn, up to 30 seconds in total. On the consumer side, Google Flow carries the model for every account, with the longer clips and extension reserved for AI Plus, Pro and Ultra subscribers.

Free and plan-included access

Gemini Omni 1.2 Flash is free to try without the API. Google Vids gives anyone with a Google account three free 15-second Omni 1.2 Flash videos a month. Google Flow hands out 50 free credits a day, and Omni Flash is free inside YouTube Shorts for creators 18 and over. The Gemini app includes it for every Google AI plan. For casual use, those paths cost nothing beyond a plan you may already have. You can see how the tiers stack up in our full Gemini pricing breakdown.

Resolution and clip limits

Gemini Omni 1.2 Flash generates at 720p at most, at 24 frames per second. It can upscale to 1080p or 4K, but an upscale is not the same as generating at the higher resolution. Clips top out at 15 seconds per generation and 30 seconds with extension. First-and-last-frame interpolation, where you supply the opening and closing images, is still capped at 10 seconds. That makes the model a fit for social clips, product spots and quick concepts rather than long-form footage.

How to Access Gemini Omni Flash

Gemini Omni 1.2 Flash is available in five places: the Gemini app, Google Flow, Google Vids, YouTube Shorts and the Gemini API. Pick the one that matches how much control you need and what you already pay for.

In the Gemini app

The Gemini app runs Omni 1.2 Flash for anyone with a Google AI plan who is 18 or older. You type a prompt, generate a clip, then keep the conversation going to refine it. Clips in the app run 3 to 10 seconds, and the 15-second length is reserved for Google AI Ultra, according to Google’s help page. Uploading your own video to edit is not available in the EEA, Switzerland, the UK and some US states.

In Google Flow

Google Flow runs Omni 1.2 Flash for every account, including free ones. Every user gets 50 Flow credits a day. A clip of up to 10 seconds costs 1 credit per second at 360p and 3 credits per second at 720p. That lets a free account make one 10-second 720p clip a day (30 credits) or five 10-second 360p drafts. Clips longer than 10 seconds, extension and upscaling are for AI Plus, Pro and Ultra subscribers, and 4K upscaling is Ultra only. Google retires Omni Flash 1.1 in Flow on December 5, 2026.

In Google Vids

Google Vids is the cheapest way to get a full 15-second clip. Since November 5, 2026, anyone with a Google account or Google Workspace subscription gets up to three free 15-second Omni 1.2 Flash videos a month, according to Google’s Google Vids announcement. You can upload reference images, and Google says Vids keeps characters and scenes consistent from the first frame to the last.

In YouTube Shorts

Google wired Omni Flash into YouTube Shorts Remix, where creators 18 and over can generate clips at no cost. It is the easiest way to test the model without a subscription, and it is aimed squarely at short vertical video, which is why 9:16 output is a first-class option.

Through the API

Developers use the Gemini API and Google AI Studio with the model ID gemini-omni-1.2-flash, on the paid tier only. Omni Flash first reached the API as a preview on June 30, 2026. Generation runs on the Interactions API and the current google-genai SDK, and you set the clip length with a duration from 3 to 15 seconds (10 by default). Google also doubled the paid-tier rate limits with 1.2. This is the path for anyone building Omni Flash into a product. VentureBeat has a useful writeup of the developer API rollout.

Gemini Omni Flash vs Veo

Gemini Omni 1.2 Flash is Google’s fast, conversational video model, while Veo 3.1 is the line built for polished, high-fidelity output. With 1.2, Omni Flash now also makes the longer clips. The table below shows where each one wins, with Flow credit costs from Google’s Flow help page.

If you want speed, iteration, and a low bill, Omni Flash is the pick. If you need broadcast-grade resolution, Veo is the better tool. In Flow that now mostly means Veo 3.1 Lite, because Google retires Veo 3.1 Fast and Quality there on December 5, 2026. Many creators will use both, roughing out ideas in Omni Flash and finishing hero shots in Veo, and for a non-Google option our guide to Meta’s AI video generator covers another fast route to short clips.

Pairing Omni Flash With Nano Banana 2 Lite

Google designed Omni Flash to work alongside Nano Banana 2 Lite, its fastest image model. Nano Banana 2 Lite generates a still in about four seconds for roughly $0.034 per 1K-resolution image. That makes it ideal for locking down a look before you spend on video. And if you want to storyboard for free first, the base model has a no-cost tier that we cover in Is Nano Banana free?.

The workflow is straightforward. You generate a clean reference image with Nano Banana 2 Lite, hand that image to Omni Flash, and ask the model to animate it into a moving clip. Starting from a fixed still gives you far more control over composition and character than describing everything from scratch, and it keeps your costs down because the expensive step only runs once you like the frame. Omni 1.2 Flash accepts up to six reference images per request, so you can pass a character, a product and a style frame together.

If image generation is your main interest, our roundup of the best AI models and our guide to Seedream 5 cover the fastest options across providers.

The Limits You Should Know

Gemini Omni 1.2 Flash is fast and cheap, but it is not magic, and Google is upfront about the rough edges. Knowing them upfront saves you wasted credits.

Generation still caps at 720p, and an upscale to 1080p or 4K is not the same as generating at the higher resolution. Omni Flash is still the wrong choice for anything that has to look crisp on a large screen. Google says fast sports action and complex physics still need work. The API does not accept uploaded audio or audio references, and any audio inside a reference video is ignored. Editing or extending a video you upload yourself is not available in the EEA, Switzerland or the UK. Google has only evaluated English prompts, so other languages may give uneven results.

None of that makes it a bad model. It just means Omni Flash is best treated as a rapid idea and social-clip engine rather than a full replacement for a video editor or a premium generator.

Where Fello AI Fits

The cheapest Omni Flash clip is the one you only generate once, so most of the work happens before the video model runs. With Fello AI you write the script and the shot list with GPT, Claude or Gemini, then generate the reference frame in the same app. You don’t open a second subscription for each step. Fello AI is a native Mac, iPhone and iPad app that puts those models and GPT Image generation behind one plan.

When the frame looks right, hand it to Omni Flash and spend your credits on the clip, not on guesswork. You can grab it on the Fello AI App Store page and see how it compares in our look at whether Gemini is free.

The Bottom Line

Gemini Omni 1.2 Flash makes AI video cheaper and longer: $0.07 per second at 720p, clips up to 15 seconds, and edits by plain description. Start with the three free clips a month in Google Vids or the daily credits in Google Flow. Move to the API once you know Omni Flash fits your workflow. Google’s text-side Flash line moved on as well, and our Gemini 3.6 Flash breakdown covers the July 2026 update.