Nano Banana 2.1: How to Use Google’s New Image Model, Pricing and vs Nano Banana Pro

Nano Banana 2.1 is Google’s newest everyday image model, released on 6 October 2026 as an update to Nano Banana 2. It promises better image quality, sharper text inside images, more consistent characters across edits and support for very wide or tall formats, while Google has roughly halved the per-image API price at 1K and 2K resolution. This guide explains what changed, how to use Nano Banana 2.1 in the Gemini app and through the API, what it costs, and when Nano Banana Pro is still the better choice.

Everything below was checked on 8 October 2026 against Google’s Gemini API changelog, pricing page and image generation guide, the Google DeepMind model card for Nano Banana 2.1, and Gemini’s own help pages. Google is rolling the model out gradually, so what you see in your account may differ for a few days.

What Is Nano Banana 2.1?

“Nano Banana” is Google’s friendly name for the image generation and editing models inside Gemini. Nano Banana 2.1 is the new version of the mid-range model: faster and cheaper than Nano Banana Pro, but more capable than the budget Nano Banana 2 Lite. According to its model card, it is based on Gemini 3.6 Flash, accepts text and images as input, and returns images and text.

What’s New in Nano Banana 2.1?

Google’s changelog lists five main improvements over Nano Banana 2:

• Better visual quality at 1K, 2K and 4K.
• Stronger prompt adherence, so the image matches more of what you asked for.
• Multi-turn character consistency: people and objects keep their look across several rounds of edits.
• Improved text rendering for posters, menus, diagrams and infographics.
• Wide and panoramic formats: very wide or tall aspect ratios now render more cleanly.

The DeepMind model card backs this with Google’s own human-preference tests. With thinking enabled, Nano Banana 2.1 scored an Elo of 1,050 for overall text-to-image preference, against 990 for Nano Banana 2 and 935 for Nano Banana Pro. On infographic factuality it scored 0.521, compared with 0.265 for Nano Banana Pro. These are Google’s figures, not independent tests, but they suggest the new model is a genuine upgrade rather than a minor patch.

The model card is also frank about weaknesses. Small text, long paragraphs and full-page text can still render poorly, characters do not always match their reference photos exactly, mask-based edits sometimes ignore part of the instruction, and the model occasionally confuses left and right.

How to Use Nano Banana 2.1 in the Gemini App

For most people, the Gemini app is the easiest way in, and image generation is available on the free plan. Reports describe a gradual rollout to the app, and on 8 October Gemini’s help pages still described the app’s image tool as Nano Banana 2, so you may not see the “2.1” label straight away.

1. Open Gemini at gemini.google.com or in the Gemini app on Android or iPhone, and sign in with your Google Account.
2. Choose the image tool. On the web, open the sidebar and select Images; in the chat box you can also open the tools menu and pick Create images.
3. Write a detailed prompt. Describe the subject, action, setting, style, lighting and aspect ratio. For example: “A 16:9 poster for a Dublin coffee shop’s autumn menu, warm lighting, the words ‘Pumpkin Spice Week’ in bold serif type at the top.”
4. Edit with follow-up messages. Ask Gemini to change the background, swap an object or fix the text. Each follow-up keeps working on the same image, which is where the improved character consistency helps.
5. Use your own photos. Upload one or more images and describe what to combine or change.
6. Download the result. Hover over the image and choose Download full size. Gemini’s help page says downloads are 1K without a Google AI plan and 2K with one.

A few rules apply. You must be signed in, editing images requires you to be 18 or over on a personal account, and image generation counts against Gemini’s usage limits, which Google says refresh every five hours up to a weekly cap. Paid Google AI Plus, Pro and Ultra subscribers get higher limits and can redo an image with Nano Banana Pro for more detail. If you already use Gemini for repeatable tasks, our guide to creating Gemini Skills shows how to save prompts you use often.

Other places to find it

According to the model card, Nano Banana 2.1 is also coming to AI Mode in Google Search, Google Flow (Google’s AI filmmaking tool), Google Stitch (its UI design tool) and Google Ads. Availability in each product depends on Google’s rollout.

How to Use the Nano Banana 2.1 API

Developers can try prompts in Google AI Studio and then call the model from code. Note that the Gemini API itself has no free tier for Nano Banana 2.1, so you need billing enabled on your project.

Step 1: Get an API key

Sign in to Google AI Studio, create an API key and enable billing on the linked Google Cloud project. Store the key in an environment variable called GEMINI_API_KEY rather than pasting it into your code.

Step 2: Install the SDK

pip install google-genai pillow

Step 3: Generate your first image

Google’s image generation guide uses the Interactions API. This minimal example creates an image and saves it to disk:

from google import genai
import base64

client = genai.Client()

interaction = client.interactions.create(
    model="gemini-nano-banana-2.1",
    input="A flat-lay photo of a London street-food menu, with the title 'Borough Bites' in clean sans-serif type",
)

with open("menu.png", "wb") as f:
    f.write(base64.b64decode(interaction.output_image.data))

Step 4: Control size, shape and thinking

• Resolution and aspect ratio are set with a response_format object, for example {"type": "image", "aspect_ratio": "16:9", "image_size": "2K"}. Use an uppercase “K”; Google says lowercase values are rejected.
• Thinking level can be minimal, medium or high. Medium is the default, thinking cannot be switched off, and thinking tokens are billed.
• Editing: send your own image (base64-encoded, with its MIME type) together with a text instruction.
• Multi-turn editing: pass the previous_interaction_id with a new instruction, such as “translate the text into German but keep the layout”.
• Grounding: add the google_search tool for current information such as weather or live data. If you use it, Google requires you to display the search suggestions it returns.

Step 5: Migrate from Nano Banana 2

If you already use gemini-3.1-flash-image, switch the model ID to gemini-nano-banana-2.1. Google’s changelog marks the old model as deprecated but, when we checked, said no shutdown date had been announced; some third-party reports give an earlier date, so watch the changelog. Move any 512px requests to 1K (or to Nano Banana 2 Lite), set the thinking level explicitly, and recheck budgets for prompts that use many reference images, because input tokens now cost more.

Nano Banana 2.1 Pricing

Here are the standard Gemini API prices from Google’s pricing page, in US dollars, compared with the model it replaces:

The Batch API halves the image prices for jobs that do not need an instant answer: $0.0168 for a 1K image, $0.0252 for 2K and $0.0567 for 4K. Grounding with Google Search includes 5,000 free requests a month, then costs $14 per 1,000 requests.

Two things are worth noting. First, some early news reports quoted a lower 4K price than the $0.113 shown on Google’s pricing page when we checked, so always confirm the live page before budgeting. Second, the cheaper image output is partly offset by input that costs three times as much as before, plus billed thinking tokens. If your workflow sends many reference images per request, test the real cost on a small batch first.

Prices are listed in dollars. UK and EU customers are billed through Google Cloud, where local currency and VAT may apply at checkout.

Nano Banana 2.1 vs Nano Banana Pro

“Nano Banana 2.1 vs Pro” is one of the first questions people are searching for, in English and in French and German too. Google now offers three main image models:

Choose Nano Banana 2.1 for most jobs: social graphics, product mock-ups, thumbnails, infographics and iterative edits. It costs about a quarter of Pro at 1K and, on Google’s own preference tests, it actually scored higher overall. Choose Nano Banana Pro when you need its separate style reference images or more character references, or when Pro gives better results for your specific prompts. Choose Lite for high-volume, simple images where speed matters more than detail.

How does it compare with rivals outside Google? Midjourney remains popular for artistic styles, and ComfyUI gives you full local control over open models. Our guides on getting started with Midjourney and setting up ComfyUI cover those options, and our ChatGPT vs Gemini vs Claude comparison looks at the wider assistants.

Practical Use Cases

• Marketing visuals: generate ad variations in several aspect ratios, from square posts to very wide web banners.
• Infographics and diagrams: the improved text rendering and search grounding make simple explainers more accurate, though you should still check every figure.
• Product photography: combine up to ten object references to place products in new scenes.
• Consistent characters: keep a mascot or presenter looking the same across a series of images.
• Video thumbnails: create and refine YouTube thumbnails quickly; our YouTube automation workflow shows where this fits in a production pipeline.
• Translation of visuals: ask for the same poster in English, German, French or Polish while keeping the design.

Tips for Better Results

1. Be specific. Google’s guide stresses that more detail gives more control. Name the style, camera angle, lighting and layout.
2. Spell out any text exactly, including where it should appear and the type of font.
3. Use higher resolution for text-heavy images. The model card warns that small text is the weakest area.
4. Ask it to use search when the image depends on real-world facts or current data.
5. Edit in small steps. Several short follow-up instructions usually work better than one long list of changes.
6. Check rights and disclosure. Every image carries a SynthID watermark; be transparent when you publish AI-generated visuals.

Pros and Cons

Pros

• Lower per-image prices than Nano Banana 2 at every resolution
• Better text rendering, prompt adherence and character consistency, according to Google
• Very wide and tall aspect ratios
• Available in the free Gemini app as well as the API

Cons

• No free API tier and higher input token prices
• 512px output removed
• Gradual rollout, so the app may still show the older model
• Small text and exact likeness can still be unreliable

Frequently Asked Questions

Is Nano Banana 2.1 free?

You can generate images in the Gemini app on the free plan, within usage limits. The Gemini API has no free tier for Nano Banana 2.1, so developers pay per image.

When was Nano Banana 2.1 released?

Google’s Gemini API changelog lists Nano Banana 2.1 as generally available on 6 October 2026, with a gradual rollout to Google’s apps.

What is the difference between Nano Banana 2.1 and Nano Banana Pro?

Nano Banana 2.1 is the faster, cheaper everyday model; Nano Banana Pro is Google’s premium image model with style references and more character references. At 1K, Pro costs about four times as much.

How do I use Nano Banana 2.1 in Gemini?

Open Gemini, choose Images or Create images, and describe the picture you want. Follow-up messages edit the same image. The new model is rolling out gradually.

Does Nano Banana 2.1 add a watermark?

Yes. Google says every generated image contains an invisible SynthID watermark that identifies it as AI-generated.

Is Nano Banana 2.1 available in the UK and Europe?

Google says image generation is available wherever the Gemini app is offered, and the API is available to Google Cloud customers. Check your account, as rollouts can reach countries at different times.

Conclusion

Nano Banana 2.1 is a meaningful upgrade to Google’s everyday image model: clearer text, steadier characters, wider formats and lower per-image prices. For most users, the free Gemini app is the place to start, while developers should switch to gemini-nano-banana-2.1 and recheck costs before the old model is retired. Keep Nano Banana Pro for the few jobs that need its extra references, and always check text and facts in the final image.

Sources: Gemini API changelog; Gemini API pricing; Gemini API image generation guide; Google DeepMind – Nano Banana 2.1 model card; Gemini Apps Help – generate images; Gemini Apps Help – usage limits; Notebookcheck.