Video

How to Use Gemini Omni in Google Flow: Access, Credits, Settings, and Export

Q: Credits And Generation Costs?

The most important cost habit is simple: credit costs are per generation, not merely per request. Flow Help notes that some requests can create multiple generations. That can make one exploratory prompt more expensive than expected if you are not watching the settings.

By Sagnik Bhattacharya 20 May 2026 7 min read

Coding Liquids blog cover featuring Sagnik Bhattacharya for using Gemini Omni in Google Flow, with model settings, credit counters, asset panels, and export controls.

Google Flow is the most practical place to organise Gemini Omni video work because it gives you a project surface, model settings, credits, assets, and edit operations in one workflow. The key is to treat Flow like a production board, not just a prompt box.

I teach Flutter and Excel with AI — explore my courses if you want structured learning.

This guide is the access and settings companion to the beginner Gemini Omni tutorial. Once your Flow setup is clear, use the prompt guide and the editing guide to improve output quality.

Note: Fact check current as of 20 May 2026: Google's Gemini Omni launch announcement says Gemini Omni Flash is rolling out to Google AI Plus, Pro, and Ultra subscribers globally through the Gemini app and Google Flow. Costs, feature support, and geography can change, so confirm the latest availability and credit cost inside the live Flow settings panel.
Note: Free Flow credits are officially described as usable for Veo 3.1 Lite, Fast, and Quality generations, not Gemini Omni Flash.

Quick answer

Open Google Flow, check whether Gemini Omni Flash is available in your account, choose the active model and supported duration, confirm the credit cost before generating, keep your assets organised in a project, then export only after you have reviewed motion, text, audio, and provenance.

You want the current access, credits, and settings picture before starting a Gemini Omni batch.
You need a repeatable way to choose a short test duration, review the result, and log credit spend.
You want to avoid burning credits on avoidable prompt and settings mistakes.

Access And Subscription Requirements

Google's Gemini Omni launch announcement says Gemini Omni Flash is rolling out to Google AI Plus, Pro, and Ultra subscribers through the Gemini app and Google Flow. That means you should check access inside your own account before planning a paid production workflow or promising a client that the model will be available on a free account.

Flow's broader video-generation surface includes multiple models. If a feature is not supported by the active model, Flow documentation says the interface can notify you. For practical work, always read the active model label and the cost shown in settings before clicking generate.

Account or model area	Officially relevant detail	Practical takeaway
Free Flow credits	Listed for Veo 3.1 Lite, Fast, and Quality generations.	Useful for exploring Flow, but do not assume Gemini Omni Flash is included.
Gemini Omni Flash	Listed for Plus, Pro, and Ultra subscribers.	Check your subscription before planning Omni work.
Google AI Pro	Flow Help lists 1,000 monthly Google Flow credits.	Enough for controlled testing if you manage batches carefully.
Google AI Ultra tiers	Flow Help lists higher monthly credit allowances.	Better for larger production batches and 4K upscaling needs.

Credits And Generation Costs

The useful cost habit is not memorising a number from an old screenshot. The useful habit is checking the credit cost shown in Flow before generation and saving it beside the prompt. Flow Help also warns that costs can be per generation and that some requests can create more than one generation, so one prompt can cost more than you expected.

Use this cost log for every serious test. It turns Flow from a creative playground into a workflow you can budget.

Field to record	Why it matters
Account plan and starting credits	Shows whether the test was run on Plus, Pro, Ultra, or another eligible plan.
Active model shown in settings	Prevents accidental comparisons between different model surfaces.
Duration and aspect ratio	Lets you compare only like-for-like outputs.
Credit cost shown before generation	Uses the live Flow value instead of stale pricing.
Number of outputs created	Some requests may create more than one generation.
Usable output decision	Lets you calculate cost per usable clip, not just cost per attempt.

Settings To Check Before Generation

Flow settings should become part of your prompt habit. A good prompt can still waste credits if the active model, aspect ratio, duration, or reference mode is wrong.

Use this pre-flight check before every meaningful batch.

Active model: confirm Gemini Omni Flash when that is the model you intend to test.
Duration: choose the shortest duration that can prove the shot, then increase only when the story needs it.
Aspect ratio: portrait for Shorts and Reels, landscape for decks, websites, and YouTube landscape.
References: attach only the assets that matter to the current shot.
Credit estimate: screenshot or note the shown cost for benchmark records.
Upscaling: do it at the end, not before you know the clip is usable.

A Practical Flow Project Structure

Flow projects become messy quickly if you generate lots of variations. Name the project after the campaign or video, then keep prompt versions and assets together. This makes your later review much less painful.

For a more developer-friendly version of the same idea, use the VS Code Gemini Omni workflow to store prompts, manifests, asset notes, and evaluation rubrics outside Flow.

Create one Flow project per campaign, product, or video series.
Name prompts with the pattern scene-number_goal_version.
Save the exact prompt beside the exported clip name.
Keep one review sheet with scores for prompt adherence, motion stability, text, audio, consistency, editability, and credit cost.
Do not upscale or export final files until the review score says the clip is usable.

Export And Review Workflow

Export is not the end of the process. AI video often needs trimming, captions, sound balancing, brand checks, and rights review before publishing. For Shorts, read the Gemini Omni YouTube Shorts guide before uploading or remixing.

When you export, preserve the working history: prompt, model, date, duration, aspect ratio, references, and edit chain. That record is useful for repeatability and for answering stakeholder questions later.

Step-by-step: create a Gemini Omni project in Flow

Use Flow like a production workspace, not a random prompt box. The exact models and credit costs can change, so check the active model and settings inside Flow before every serious generation.

Confirm access first. Make sure your account is eligible for Flow in your region, that your age is verified, and that your Google AI plan gives you access to the feature you want to use.
Open Flow and create a project folder. Name it by campaign or topic, such as omni-shorts-q2-dashboard. Keep one project per output goal.
Check model and settings. In the prompt box settings, confirm the active model, aspect ratio, available features, and latest credit cost before you generate. Flow Help specifically advises checking the active model and costs in settings because limits can change.
Add references only when they help. Upload a product image, chart, storyboard, or base clip if preservation matters. If you only need a simple idea test, start text-only.
Paste a structured prompt. Use subject, action, camera, style, lighting, constraints, and final-frame instructions. Avoid giant paragraph prompts with ten goals.
Generate one controlled version. Do not immediately ask for four different creative directions. First prove that the subject, motion, and framing work.
Review before editing. Check prompt adherence, motion stability, text rendering, reference preservation, audio usefulness, and whether the clip can be cut into a sequence.
Make one edit request. Ask for a single change such as "keep everything the same, but make the camera locked-off" or "preserve the chart and remove the extra labels".
Export and record the run. Save the output with the prompt, date, model, aspect ratio, duration, and credit spend. This gives you a real cost per usable clip later.

Flow project checklist

Before you publish or hand off the clip, make sure the project contains enough information for someone else to reproduce or audit the result.

The final prompt and edit prompts are saved.
Reference assets are named clearly and stored beside the project.
The active model, feature, duration, aspect ratio, and generation date are recorded.
Credit cost is recorded per generation, not just per finished clip.
The exported file name describes the clip, version, and use case.
Any manual captions, edits, or disclaimers are documented outside Flow.

Troubleshooting Flow outputs

Problem	Likely cause	Next edit
Wrong aspect ratio	Settings were not checked before generation.	Set vertical or landscape explicitly and regenerate.
Subject changes identity	Reference preservation was vague.	List the exact shape, colour, logo, or character details to keep.
Generated text is messy	The prompt asked for too much small text.	Remove generated text and add captions manually.
Clip is beautiful but unusable	No final-frame or editability instruction.	Ask for stable first and final frames.
Credits disappear too quickly	Too many broad experiments.	Run short controlled tests and log cost per usable clip.

Common mistakes

Assuming free Flow credits cover Gemini Omni Flash.
Testing the longest available clip before a short prompt sanity check.
Changing model, prompt, duration, and references in the same benchmark run.
Upscaling clips that have not passed the review checklist.
Failing to save the prompt and settings beside the exported file.

Sources

These official references are useful if you need the product or framework documentation alongside this guide.

Want to create better AI content?

My courses cover practical AI workflows for content creation, video production, and marketing with real projects.

Browse courses