Image model · by Google DeepMind
Nano Banana 2 AI image generator
Nano Banana 2, also called Gemini 3.1 Flash Image, is an image model from Google DeepMind. On DreamBench it now runs on Nano Banana 2.1, Google’s October 2026 update, and makes freeform images and the sheets your videos are built from.
Available on every plan · Released February 2026
- 6 credits per image
- Up to 10 images per run
- 6 aspect ratios
- Character, prop and location sheets
- On every plan
Start with a prompt
Try it now
Edit the idea or write your own. Press Create with Nano Banana 2 and the model is already picked for you.
With any DreamBench plan
Tools that run on Nano Banana 2
Each tool takes one photo and returns one result made by Nano Banana 2, ready to download full size or to turn into a video.
AI background changer
Put the subject of any photo in a new scene, with the light to match.
Open the toolAlso: Background color changer
Strengths
What Nano Banana 2 is good at
- 01
Words you can read
Short headlines, signs and large labels come out as legible lettering, ready for a poster, a menu board or a card.
- 02
Knowledge of the world
It draws on what Gemini knows, which helps with infographics, labeled diagrams and scenes set in a particular place or period.
- 03
Characters that stay recognizable
Several characters and objects can keep their look across a set of images, which is what a storyboard or a series of product shots needs.
- 04
Detailed prompts, followed closely
Complex prompts full of specifics about subject, style and color are followed closely.
At a glance
Nano Banana 2 on DreamBench
credits per image
images per run
aspect ratios
1:1, 4:3, 3:4, 16:9, 9:16 and 21:9
things to make
Freeform images, character sheets, prop sheets, location sheets and location schematics
The model
What is Nano Banana 2?
- Made by Google DeepMind
- Released February 2026
- Image model
Nano Banana 2 is an image model from Google DeepMind, released on February 26, 2026. Its technical name is Gemini 3.1 Flash Image, and it is built on Gemini 3 Flash, drawing on Gemini’s knowledge of the world when it makes infographics, diagrams or historical scenes.
Google built it to draw readable text into images, from posters and mock-ups to greeting cards, and to keep several characters and objects recognizable across a set of images. It follows complex, detailed prompts closely.
Google is open about its limits: the model can still get spelling, small faces and fine detail wrong, and it can get facts wrong too, so check any text or data before you publish. Every image carries SynthID, Google’s invisible watermark that marks it as AI-generated.
In October 2026 Google released Nano Banana 2.1 as an update to Nano Banana 2, with better image quality, closer prompt following, cleaner text and characters that hold up across edits. It marked Nano Banana 2 deprecated in the Gemini API, with no shutdown date announced, and DreamBench now makes every Nano Banana 2 image on 2.1. Google also offers a smaller Nano Banana 2 Lite, released in June 2026; DreamBench doesn’t run it.
More from Google DeepMind: Nano Banana 2 on Google DeepMind
The details
Nano Banana 2 specs and credits
What you get when you make images with it on DreamBench.
| Made by | Google DeepMind |
|---|---|
| Credits | 6 per image |
| Images per run | 1 to 10 |
| Aspect ratios | 1:1, 4:3, 3:4, 16:9, 9:16 and 21:9 |
| What you can make | Freeform images, character sheets, prop sheets, location sheets and location schematics |
| Reference images | Up to 8 per run |
| Plans | Every plan |
Good to know
- Picking Nano Banana 2 on DreamBench means Nano Banana 2.1, Google’s model gemini-nano-banana-2.1, for every image in the run.
- DreamBench uses the model’s default image size, at which Google says small text can come out blurry. Keep lettering short and large.
- In a freeform image, put the words you want to see in double quotation marks: DreamBench passes them letter for letter to every version of the run.
- In a freeform run, each version after the first takes its own direction, such as another angle, light or palette, while keeping everything your brief states. Nano Banana 2 makes the versions one after another, and each lands in your library as soon as it’s ready.
- In Google’s own apps and API, Nano Banana 2 can also edit existing images and translate the text in them, look up pictures on the web as it works and make larger images. On DreamBench, it makes new images from your brief and your reference images.
Estimate your credits
A Nano Banana 2 run, as the composer prices it.
24credits
6 credits per image. Starter’s 1,100 monthly credits make about 183 images. You always see the exact cost before you generate.
Side by side
Nano Banana 2 compared
Nano Banana 2 beside the models people weigh it against, as each runs on DreamBench. Every cell comes from the same tables as the spec sheet above.
| Model | Images per run | Aspect ratios | What it makes | Reference images | Plans |
|---|---|---|---|---|---|
| Nano Banana 2Google DeepMind | 1 to 10 per run | 1:1, 4:3, 3:4, 16:9, 9:16, 21:9 | Freeform images, character sheets, prop sheets, location sheets and location schematics | Up to 8 | Every plan |
| GPT Image 2OpenAI | 1 to 10 per run | 1:1, 4:3, 3:4, 16:9, 9:16, 21:9 | Freeform images, character sheets, prop sheets, location sheets and location schematics | Up to 8 | Every plan |
| Seedream 5.0ByteDance Seed | 1 to 10 per run | 1:1, 4:3, 3:4, 16:9, 9:16, 21:9 | Freeform images, character sheets, prop sheets, location sheets and location schematics | Up to 8 | Every plan |
| GPT Image 2.5OpenAI | 1 to 10 per run | 1:1, 4:3, 3:4, 16:9, 9:16, 21:9 | Freeform images, character sheets, prop sheets, location sheets and location schematics | Up to 8 | Every plan |
How it works
How to make images with Nano Banana 2
Step 1
Choose Nano Banana 2
Pick Nano Banana 2 in the composer’s image mode, or press Create with Nano Banana 2 on this page.
Step 2
Describe the image
Write what you want to see, or make a character, prop or location sheet for a video you’re planning.
Step 3
Generate and reuse
Make up to 10 versions at once. Every image lands in your library, ready to use as a reference in your videos.
Prompt ideas
Prompt ideas for Nano Banana 2
Start from one of these and change it however you like.
A bakery poster with lettering
A 3:4 poster for a neighborhood bakery’s autumn opening. A crusty sourdough loaf and a few wheat stalks on a warm cream background, lit by soft window light. A large hand-lettered headline across the top reads “Fresh Out of the Oven”, with “Saturday, 8 am” in bold type underneath. Muted orange and brown palette, plenty of empty space around the bread.
One fox, 4 panels
A 4-panel comic strip in soft watercolor, 16:9. The same small ginger fox in a yellow knitted scarf appears in every panel: packing a picnic basket, cycling along a lakeside path, getting caught in a sudden shower, and drying off under an oak tree with a cup of tea. Keep the fox and the scarf identical throughout. No speech bubbles or text.
A life-cycle diagram
A clean 4:3 educational diagram of the life cycle of a frog, arranged in a circle with curved arrows: frogspawn, tadpole, tadpole with back legs, froglet and adult frog. Each stage is drawn in a friendly, flat illustration style with a large, simple label. Pale green paper background and a bold title at the top: “From Egg to Frog”.
Still have questions?
Nano Banana 2 FAQ
What is Nano Banana 2?
Nano Banana 2 is an image model from Google DeepMind, released in February 2026 under the technical name Gemini 3.1 Flash Image. It makes images from a text prompt and reference images, and Google built it to draw readable text and keep characters consistent.
Is Nano Banana 2 the same as Gemini 3.1 Flash Image?
Yes. Gemini 3.1 Flash Image is the technical name of the model Google calls Nano Banana 2, released on February 26, 2026, and built on Gemini 3 Flash. On DreamBench it makes freeform images and character, prop and location sheets, and now runs on Google’s Nano Banana 2.1 update.
Is Nano Banana 2 being replaced by Nano Banana 2.1?
Google released Nano Banana 2.1 in October 2026 as its update to Nano Banana 2, with better image quality, closer prompt following, cleaner text and characters that hold up across edits, and marked Nano Banana 2 deprecated in the Gemini API, with no shutdown date announced. DreamBench already makes every Nano Banana 2 image on 2.1.
How is Nano Banana 2 different from Nano Banana Pro?
Google built Nano Banana 2 for fast generation and close instruction following, and it replaced Nano Banana Pro as the image model in the Gemini app. Google AI Pro and Ultra subscribers keep Nano Banana Pro for specialized tasks. DreamBench doesn’t offer Nano Banana Pro.
What can I make with Nano Banana 2 on DreamBench?
Freeform images, plus character, prop and location sheets and location schematics. Sheets are saved to your library and your videos can use them as references, so the characters and places you design here can appear in your scenes.
Can Nano Banana 2 put text in images?
Yes. Google built Nano Banana 2 to draw readable text for posters, cards and mock-ups. DreamBench uses the model’s default image size, at which Google says small lettering can blur, and the model can still misspell words, so keep text short and large, put the exact words in double quotation marks, and check the image before you use it.
Do Nano Banana 2 images have a watermark?
Every image Nano Banana 2 makes carries SynthID, Google’s invisible watermark, on DreamBench or anywhere else: it is part of what the model produces, not a setting. You can’t see it in the picture, but it lets the image be identified as AI-generated. DreamBench adds no watermark of its own to images.
How much does Nano Banana 2 cost on DreamBench?
It uses credits for each image you make, and one run can make up to 10. The table on this page shows the current rate, and you always see the exact cost before you generate.
Does Nano Banana 2 cost anything to use?
Yes. On DreamBench, each Nano Banana 2 image uses credits, which come with a paid plan, and Nano Banana 2 is on every plan. The composer shows the exact cost before you generate. Google also includes Nano Banana 2 in its Gemini app and as Flow’s default image model, with the limits it sets there.
Can Nano Banana 2 keep a character consistent across images?
Yes. Google built it to keep several characters and objects recognizable across a set of images. On DreamBench, make a character sheet first, then attach it as a reference and name it with its @name in each brief; a run takes up to 8 reference images.
Can I edit my own photo with Nano Banana 2 on DreamBench?
No. On DreamBench, Nano Banana 2 makes new images from your brief and reference images. Editing an uploaded photo and translating the text in it are features of Google’s own apps and API.
How fast is Nano Banana 2?
Google built it on Gemini 3 Flash, its high-speed model line, for quick generation and edits. On DreamBench the versions of a run arrive one after another, each landing in your library as soon as it’s ready.
Should I use Nano Banana 2 or GPT Image 2 for text in images?
Both developers list text among their models’ abilities, and both note that lettering can still come out wrong. Both are in DreamBench’s image mode, as is Seedream 5.0 Pro, so run the same brief on each and keep the version that reads best; their pages are linked below.
Does DreamBench have a Nano Banana 2 API?
No. DreamBench offers Nano Banana 2 through its composer, in the browser on a phone, tablet or computer, with nothing to install. Developers can use Google’s Gemini API, linked under Sources, where the prices and limits are Google’s, not DreamBench’s.
Can I use Nano Banana 2 images commercially?
Yes. DreamBench’s Terms of Use allow personal and commercial use, including ads and client work. You need the rights to anything you upload.
Which DreamBench plans include Nano Banana 2?
Nano Banana 2 is on every DreamBench plan, from Starter at $9.99 a month. Every plan makes freeform images, character sheets, prop sheets, location sheets and location schematics with it, up to 10 images in one run.
How many images can Nano Banana 2 make in one run?
From 1 to 10 in a single run, each at 6 credits, in any of 6 aspect ratios (1:1, 4:3, 3:4, 16:9, 9:16 and 21:9). You see the exact cost before you generate.
How do I make images with Nano Banana 2 on DreamBench?
Press Create with Nano Banana 2, or pick it in the composer’s image mode, describe the image, choose how many and the aspect ratio, and generate. Add up to 8 reference images to steer the look.
Keep exploring
Other models to try
OpenAI
GPT Image 2
Put words inside the picture, from a poster headline to a speech bubble, or draw sheets for your videos.
61 credits per image
ByteDance Seed
Seedream 5.0
Make infographics, posters and photographic images, with text in many languages.
15 credits per image
Google DeepMind
Veo 3.1
Shape realistic widescreen or vertical scenes with dialogue and sound, on Veo 3.1, Fast or Lite.
Up to 8-second clips · 720p / 1080p / 4K · Sound
OpenAI
GPT Image 2.5
Make everyday images, and the character, prop and location sheets your videos are built from.
16 credits per image
Where the facts come from
Sources
- Nano Banana 2 on Google DeepMind
- Google’s Nano Banana 2 launch post
- Gemini 3.1 Flash Image model card (Google DeepMind)
- Nano Banana 2.1 in the Gemini API docs
- Gemini API changelog (Google)
Nano Banana 2 is developed by Google DeepMind. DreamBench is an independent platform that offers Nano Banana 2 alongside other models, and isn’t affiliated with or endorsed by Google DeepMind. The facts on this page were last checked on . Nano Banana 2 and Google DeepMind are trademarks of their owners, used here only to name the model and its developer.
Make images
Create withNano Banana 2
Describe the image you want, or build the characters and places for your next video.