How to Use Nano Banana for Free with Google Flow

Nano Banana is Google’s family of image-generation and image-editing models. These models are especially useful for designers: they can work with references, preserve characters and objects, change individual parts of an image, and render text inside images fairly reliably.

You do not need a subscription just to try Nano Banana. Google has an official platform, Google Flow, where you can generate images with a regular Google account.

The details matter, though: Google has changed its models and usage limits several times in recent months. Here is how it works now.

The Nano Banana model family

Google Flow currently offers three Nano Banana models. Here is how they differ.

Nano Banana Pro

Nano Banana Pro is built on Gemini 3 Pro Image. Google positions it for tasks that require complex composition, fine detail, and studio-level control. It is available by default to Google AI Ultra subscribers.

Nano Banana 2

Built on Gemini 3.1 Flash Image, this model is designed to follow complex instructions, render text more accurately, draw on knowledge from Gemini and Search, maintain consistent characters and objects, support multiple aspect ratios, and generate images from 512px to 4K.

Nano Banana 2 Lite

Built on Gemini 3.1 Flash-Lite Image, this is the fastest and most efficient model in the family. It is the default model for free Google Flow users. For most day-to-day work—covers, ad visuals, product photography, backgrounds, and concepts—Nano Banana 2 Lite is more than capable.

Can you use Nano Banana for free?

Yes, but it is important to understand how the system is set up.

Google Flow gives free users 50 Google Flow credits each day. However, according to Google’s official documentation, those credits are exclusively for video generation with Veo 3.1 Lite, Fast, and Quality. They are not used for image generation.

Image generation with Nano Banana 2 Lite is available at no charge; it is the default model in a free account. In other words, working with images through Nano Banana 2 Lite does not require a separate subscription.

That means the “20 free images a day” claim found in older material no longer describes the current system accurately. The number of available generations depends on the selected model, operation, and any limits Google applies. Check the current status directly in Flow.

How to use Nano Banana in Google Flow

The process itself is straightforward.

Step 1. Open Google Flow

Go to labs.google/fx/tools/flow and sign in with a regular Google account. You do not need a separate registration or API key.

Step 2. Create a new project

Flow is not a standard chat like Gemini. It is a workspace where you can keep generated images, videos, and different versions of assets in one project. That is useful when you are creating several visuals for one brief rather than a single random image.

Step 3. Switch to image generation

In the creation area, select image generation. Flow’s interface changes from time to time, but the workflow stays the same: project → image → model → prompt → generation.

Step 4. Check the selected model

Before generating, check which model is shown in the switcher. A free account should use Nano Banana 2 Lite by default. Do not assume that every image from Flow was made with Nano Banana Pro—first check what is selected in the interface.

Step 5. Write a prompt

The most common beginner mistake is writing something like:

That is too abstract for an AI model. It is much better to break a prompt into at least five parts: subject → composition → setting → lighting → style.

For example:

That gives the model a much more specific brief.

Which language should you use for prompts?

Google Flow officially supports Russian. At the same time, Google recommends English prompts for the most predictable results, since quality can vary with the language and complexity of a request.

In practice, I would not turn that into a rule that English is always better. You can start with Russian for a simple task. If the result is inaccurate, it helps to:

1. make the request more specific;
2. remove conflicting requirements;
3. break a complex scene into several stages;
4. then try the English version.

The problem is usually not the language—it is that the prompt is too broad.

What Nano Banana is especially useful for in a designer’s workflow

Asking AI only to “make a beautiful image” is one of the least effective ways to use it. For designers, it is much more valuable as a source of visual material for a specific brief.

1. Generate an object for a chosen composition

Instead of searching stock libraries for a suitable photo, you can create an image to fit a specific area of the layout. The difference matters: rather than settling for an image that almost works, you design one around its intended position in the banner.

For example:

The instruction about the empty left side matters more here than a long list of “beautiful” adjectives: we are designing the image for the intended composition.

2. Create a background for copy

Sometimes you already have the product but need a suitable setting. Generate only the background rather than the entire banner:

This approach is much easier to control.

3. Explore several visual directions

Need to show a client three approaches to the same product—minimalist, emotional, and premium? You do not have to build three designs straight away. Start by generating three visual directions, compare them, and only then move into the layout.

4. Edit an existing image

Nano Banana’s strength is not limited to generating from scratch. You can take an existing image and ask it to change the background, lighting, object color, individual details, or composition. This is especially useful when a photo is almost right but one particular element gets in the way.

Do not ask AI to make the entire ad banner

This is probably the most important practical point. Nano Banana handles text within images noticeably better than many older generators, but when the goal is a usable ad creative, I still prefer to split the process:

AI: object, photo, scene, background, visual idea.
Figma: composition, headline, offer, price, CTA, typography, grid, and format adaptations.

The reason is simple: the designer retains full control of the copy and structure. If the price or offer changes, you do not need to regenerate the image.

A practical Nano Banana + Figma workflow

Here is a workflow that is genuinely convenient. Say you need an ad creative for a coffee shop.

1. Choose the structure first: a large headline on the left, product on the right, a price label, and a CTA at the bottom.
2. Generate only the visual you need in Nano Banana:Iced coffee in a transparent glass, isolated composition on the right side of the frame. Warm morning sunlight, realistic condensation, premium commercial photography. Clean beige background. Leave large negative space on the left for advertising typography. No text, no logos.
3. Move the result into Figma and add the headline, price, button, typography, and brand colors.
4. Build several versions: one with a large price, one focused on the product, and one focused on the offer.

This is where AI stops being a “beautiful image generator” and becomes a practical tool in the design workflow.

If you would rather skip the step of building a banner structure from scratch, use ready-made compositions—learn more about ready-made Figma ad banner templates →

A prompt specifically for an ad creative

Instead of the broad request:

Try:

This does not merely describe an object. It accounts for the person’s placement, empty space for copy, format, lighting, and photographic style. Those details usually have the greatest influence on whether a result is useful in a real layout.

For a full walkthrough of building a banner from prompt to final layout, see how to create an ad banner in Figma →

Limits and credits in 2026

Rely on Google’s current official information, since plans and limits can change.

At the time this article was updated, free Google Flow accounts receive 50 Google Flow credits per day. According to the official documentation, these credits can only be used for video generation with Veo 3.1 Lite, Fast, and Quality. Image generation with Nano Banana 2 Lite—the default model for free users—is available at no additional charge.

The claim “Nano Banana gives exactly N free images per day” is currently inaccurate: the number of available generations depends on the selected model, operation, and service limits. Before a large task, review the current credit limits and check your account status directly in Flow.

Does Google Flow work in Kazakhstan and Russia?

Google publishes an official list of supported countries. At the time this article was updated, Kazakhstan is included among the regions where Google Flow is available, and Russian is supported in the interface and in prompts.

Russia is not listed among Flow’s currently supported countries. Availability can vary by region and account.

Is it worth starting with the free version?

If you want to:

• test image generation;
• develop a visual concept or several creative directions;
• prepare a background or product visual;
• try image editing,

then starting with free access is perfectly reasonable. A subscription becomes more useful when generation is a regular part of your process and you need higher limits, Nano Banana 2 or Nano Banana Pro, and Flow’s additional capabilities.

There is no need to buy a plan before you have tried it.

The key takeaway

In 2026, Nano Banana is more than a text-to-image generator. Its greatest value is not as a replacement for a designer, but as a tool within the design process.

The most useful sequence looks like this:

define the brief and composition → generate the visual you need → build the ad creative in Figma.

This lets you retain control of the design while greatly expanding your choice of source imagery.