Google Opens Gemini Omni to Everyone: A Step-by-Step Guide to Creating Free Videos in Vids

What is Google Vids 

Google Vids is a video editor in the Google Workspace suite. It runs in the browser and follows the same logic as Google Docs or Slides: the project is stored on Google Drive, can be opened via a link, and edited collaboratively with colleagues.

Google originally positioned Vids as a tool for video presentations. It includes templates, a timeline, a library of stock video and music, screen and camera recording, AI avatars, and AI voiceovers. The Slides to Vids feature turns a Google Slides presentation into a video with voiceover narration.

Until September, video generation in Vids required a paid Workspace subscription or a Google AI plan. Now Google has opened it up to everyone, using Omni 1.1 Flash as the model.

What Gemini Omni 1.1 Flash can do

Gemini Omni is a family of multimodal generative models from Google DeepMind. On the model page, Google describes its mission as “to create anything from anything, starting with video.” Omni takes text, images, video, audio, and drawings as inputs, and outputs a video with sound — voice, ambient noise, and music.

Google released version 1.1 Flash for developers on August 27, 2026. Key changes compared with the first version:

1. Scene extension. The model analyzes up to 10 seconds of the previous video — earlier models only considered the last second. You can extend a clip in 10-second increments up to a total length of 40 seconds.
2. First and last frame. Omni generates a continuous video between two specified frames — enabling camera fly-throughs, transitions, and looping clips.
3. 360p drafts. Low-resolution previews generate up to 60% faster and cost three times less than standard 720p.
4. Upscale to 4K. Final clips can be rendered in 1080p or 4K.
5. Video as a reference. You can add up to three seconds of video to a prompt to preserve a character’s appearance and the visual context.

Among customers already using Omni Flash via the API, Google names Adobe Firefly, Figma Weave, Runway, and GMI Cloud. Google AI Plus, Pro, and Ultra subscribers got Omni 1.1 in Google Flow on announcement day.

Only some of these capabilities made it into Google Vids.

Which Omni Features Are Available in Google Vids

In its announcement, Google highlights three creative tools, while the Vids help page describes four AI video modes.

Create. Generate a clip from a text prompt. You can attach up to seven reference images (the Ingredients section) or an AI avatar to the prompt.

Animate. Bring a uploaded JPG or PNG photo to life.

Edit. Edit an uploaded video. If the video is longer than 10 seconds, the model will process only the last 10 seconds. You can add up to three reference images.

Extend. Extend a clip created in the current session while preserving lighting, environment, and the characters’ appearance.

All modes support two aspect ratios: horizontal 16:9 and vertical 9:16. Clip length is set manually — from 3 to 10 seconds. Frame rate is 24 fps, and resolution is 720p or 1080p. Finished AI clips can be upscaled to 1080p.

Below is a comparison of Omni 1.1 Flash capabilities in Google Vids versus the Gemini API.

Vids doesn’t replace the developer API, but it does cover a use case where Omni previously wasn’t available for free: generate several scenes and immediately assemble them into a video with captions, music, and voiceover in a single window. 

Practice: a Video for Incrypted+ 

In its announcement, Google suggests several use cases: a product launch using a ready-made template, a promo for a local business based on product photos, videos for corporate landing pages, and community event announcements. We combined the first two and made a vertical ad video for Incrypted+ — the editorial team’s private crypto community.

The script follows the classic “problem — solution — call to action” structure and is based on key points from the Incrypted+ landing page:

• problem: a trader is drowning in notifications from a dozen channels that repeat the same information;
• transition: he joins the community’s voice call;
• solution: a team of analysts breaks down the market — the Incrypted+ logo glows on the wall;
• call to action: a monitor with the subscription page open next to a cup of coffee.

The final video is about 30 seconds long in a 9:16 format. 

Step 1. Open Google Vids 

Go to vids.new in a desktop browser and sign in to your Google account. A start menu will open with options for creating a video: a blank project, templates, recording, and AI generation. 

Select Create AI videos. A side panel will open on the right with Create, Animate, Edit, and Extend modes, and below the prompt input field — aspect ratio and duration toggles. 

Before your first generation, it’s worth checking that the project canvas format is vertical. By default, the clip inherits the canvas format, but the Portrait toggle under the prompt lets you set it manually. 

Step 2. The first scene: the problem 

The first clip sets the main character and the overall mood of the video. Build the prompt using Google’s help recommendation: subject, location, action, camera, lighting, and sound. Set the duration to 8 seconds, and the format to Portrait. 

Prompt

Vertical shot, night. A young crypto trader in a grey hoodie sits in a dark room lit only by three monitors with flickering red candlestick charts. His phone on the desk keeps buzzing with endless notifications from dozens of chat channels. He rubs his eyes, overwhelmed and tired. Handheld camera, slow push-in on his face, cold blue light, tense ambient sound, constant notification pings.

Click Generate. Generation takes anywhere from one to several minutes, after which the clip appears in the side panel.

If you’re happy with the result, click Insert — the clip will be added to the project timeline. Without this step, it will remain only in the session history. 

Step 3. Extending the scene 

Extend mode is the key new feature of Omni 1.1 in Vids. The model continues the clip from where it ended, while preserving the character, lighting, and setting. Select the first clip in the sidebar and click Extend beneath it.

In the prompt, describe only what happens next. There’s no need to repeat the scene description — the model takes the context from the original clip.

Prompt

Then he pushes the phone away, puts on headphones and joins a live voice call. On the main monitor a calm, well-organized community workspace appears with separate channels for trading, airdrops and DeFi. His face relaxes, the cold blue light slowly turns into warm orange. The camera gently pulls back.

It’s worth paying attention to the continuity: the hoodie, hairstyle, and monitor placement are preserved. This consistency is what Google calls the key advantage of Omni 1.1 — the model now analyzes up to 10 seconds of the previous video, rather than just the last frame.

Step 4. A scene with the team and logo via Ingredients 

For the third scene, you need the Incrypted+ logo. Omni doesn’t know what it looks like, so we provide it as a reference image. In the Create tab, click Ingredients and upload a PNG with the logo. 

In the prompt, explicitly specify where to place the logo. Duration — 8 seconds.

Prompt

A small team of crypto analysts works late at night in a modern studio with a brick wall. They discuss market charts on a large wall screen; one analyst points at a rising green chart and the others nod. The attached logo glows as a neon sign on the brick wall behind them. Smooth gimbal orbit around the team, cinematic lighting, subtle electronic music.

Step 5. Bringing the photo to life 

The final scene is based on Google’s “photograph a product and turn it into a promo” script from the announcement. We photograph a monitor with the Incrypted+ page open.

Switch to the Animate tab and upload the photo via Add image. Google’s help docs recommend describing not the scene, but the motion in the prompt — the model can already see the scene in the shot.

Prompt

Slow camera push-in toward the monitor screen. Soft morning sunlight sweeps across the desk, steam gently rises from the coffee cup. Calm, confident mood, no other movement.

If the photo’s aspect ratio differs from 9:16, Vids will add black bars. To avoid that, you need to shoot in vertical format from the start. 

Step 6. Editing and voiceover

Four clips on the timeline give you about 30 seconds of video. From there, we work in the standard Vids editor, without any further generations.

Add a voice-over.

Prompt

Tired of looking for alpha in dozens of identical channels? At Incrypted+, the market is analyzed by a team: voices every week, trading, airdrops and DeFi, support in Discord around the clock. Join the community on Incrypted.

If the generated scenes have their own audio, it’s worth turning the volume down on the timeline so it doesn’t drown out the voice. You can grab background music from Vids’ built-in stock library.

Step 7. Export

Download the finished video via File → Download in MP4 format. The project stays on Google Drive, so you can come back to it for edits.

In the end, the video took four generations out of six, plus one credit for the voice-over. All clips received an invisible SynthID watermark.

Omni Limitations in Google Vids

Free access in Vids removes the cost barrier, but it doesn’t eliminate the limitations of the model and the platform itself. Before planning work in Vids, it’s worth keeping the following in mind.

Small quota. Six generations per month is basically one short video. Any failed attempt eats up a noticeable share of the limit, and Vids doesn’t offer quick 360p drafts like the API does.

Short clips. The maximum length of a single clip is 10 seconds. Longer scenes have to be assembled from extensions, each of which consumes your limit.

Use outside Vids. Google’s help docs state that generated images and videos are intended for use only within Vids. Before publishing a video on social media or using it in ads, you should review the service’s terms of use.

Content rules. Generation is governed by Google’s generative AI prohibited use policy: deepfakes, content created without the consent of the people depicted, and illegal materials are prohibited.

Watermark. Each clip receives an invisible SynthID marker. You can’t remove it, and any viewer can verify the video’s origin.

English language. Help me create and some other AI features work only in English.

Most of these limitations are removed with a paid plan or by switching to the API, but the free limit is enough for a first look at the model.

Who Omni in Google Vids is for

Free Omni in Vids is, first and foremost, an entry point. Previously, to try Google’s video model, you needed a Google AI subscription. Now a regular Gmail account is enough, and the generated scenes go straight into the editing timeline with captions, music, and voiceover.

This format works well for small projects and creators who need one or two short videos a month: an announcement, a product promo, or an event invite. For regular video production, six generations isn’t enough — you’ll need Google AI Pro with 500 seconds per month, or work via the Gemini API with extensions up to 40 seconds and upscaling to 4K.

For Google, launching Vids is a way to bring Workspace users into its ecosystem — people who previously chose other services for short videos. The company has already outlined the next step: voiceover powered by Gemini 3.8 Flash-Lite should turn Vids into an end-to-end tool where the script, visuals, and voice are created in a single window.