Nano Banana 2.1

Name: Gemini Image
Author: Google DeepMind

Nano Banana
Nano Banana 2
Nano Banana 2.1
Nano Banana Pro

byGoogle DeepMind

Google's newest Nano Banana renders sharper text and holds faces steadier. Fuse up to 10 references and retouch a photo by saying what to change.

Direct the frame

How Gemini Image works

You work in a back-and-forth: hand over references with roles, say what you want in full sentences, and refine the result the way you'd brief a retoucher.

Upload your references

Add up to 10 images and name each one's role: the character, the garment, the room, or the look to borrow.

Direct it like a creative director

Write full sentences for a new scene, or state one change and a keep-list for an edit, such as 'change only the jacket; keep face and background'.

Pick a tier and resolution

Draft on Nano Banana 2.1 at 1K, then re-render keepers at 2K or 4K, or move the densest layouts to Nano Banana Pro.

What Gemini Image is good at

Because the image comes from a language model rather than a separate diffusion pipeline, it follows instructions closely, keeps people and products consistent, and writes text you can read. Nano Banana 2.1, the default, improves all three.

Multi-reference identity fusion

Combine up to 10 reference images so one person, product or wardrobe carries across angles and settings. Google rates Nano Banana 2.1 for up to 4 consistent characters in a single image; name each reference's role to keep identity from drifting.

Edits that leave the rest alone

Say what to change and what to keep, and the edit blends into the source's lighting and grain instead of looking pasted on. Swap a jacket, relight a street, or move a character to night.

Legible type on the default tier

Nano Banana 2.1 brings sharper text rendering and more accurate infographic layouts to everyday drafts. Put the words in quotation marks and describe font and placement; save Nano Banana Pro for the densest layouts.

Half-price drafts, up to 4K

Nano Banana 2.1 costs about half what Nano Banana 2 did at 1K and 2K, with better realism at every resolution. Iterate at 1K, then finalize at 2K or 4K without switching models.

Made with Gemini Image

A spread of clean, photoreal frames across interiors, fashion, documentary and product work, each naming a lens, film stock or material to avoid the default stock look.

Sculptural interior visualization

Cinematic night portrait at a lone petrol pump

Luxury brand product with high-contrast palette

Tabletop commercial still

What people build with Gemini Image

Fashion teams, storyboard artists, poster designers and product studios use it when a set of images has to feel like one project.

Fashion lookbooks and styling

Feed in up to 10 reference angles and keep a model's face, wardrobe and styling consistent across an entire collection, then restyle a single garment without redoing the shoot.

Cinematic storyboard frames

Build storyboard frames with a recurring cast, then change weather, props or lighting in plain language while the composition holds. 'Same character, now seated at night' is a valid instruction.

Posters and album sleeves

Put real words on the artwork: titles, labels and small credits. Draft the layout on a Flash tier, then re-render the keeper on Pro for dense text and print-ready size.

Product design and mockups

Drop a product into new contexts from reference photos, or imagine a prototype with believable materials and studio light, keeping the object consistent across a set.

Search-grounded factual stills

Turn on search grounding for landmarks, weather, products or data-driven infographics so the image reflects the real world. Leave it off for pure fiction.

Gemini Image specs

Plans and credits

Versions: Nano Banana (Gemini 2.5 Flash), Nano Banana 2 (Gemini 3.1 Flash), Nano Banana 2.1, Nano Banana 2 Lite (Gemini 3.1 Flash Lite), Nano Banana Pro (Gemini 3 Pro)
Inputs: Prompt (text), System Prompt (text), Image (up to 10)
Output: image
Resolution: 1K, 2K, 4K
Aspect ratios: Square (1:1), Landscape (4:3), Portrait (3:4), Landscape (3:2), Portrait (2:3), Landscape (5:4), Portrait (4:5), Landscape (16:9), Portrait (9:16), Ultrawide (21:9)
Cost: 46–364 credits per run

Gemini Image questions

Yes. Nano Banana is the nickname for Google's Gemini Image family, and this node covers its tiers: Nano Banana (2.5 Flash), Nano Banana 2 (3.1 Flash), Nano Banana 2 Lite, Nano Banana Pro (3 Pro), and a Nano Banana 2.1 refresh that is the default.

Google released Nano Banana 2.1 on 6 October 2026 as the successor to Nano Banana 2 (Gemini 3.1 Flash Image). It improves visual quality and realism at 1K, 2K and 4K, renders text and infographic layouts more accurately, and holds characters more consistently. It also costs about half as much as Nano Banana 2 at 1K and 2K. Nano Banana Pro remains the pick for the densest text and most complex layouts.

Yes. Google deprecated Nano Banana 2 (gemini-3.1-flash-image) on 6 October 2026, will shut it down on 29 October 2026, and names Nano Banana 2.1 as its replacement. New Gemini Image nodes already default to Nano Banana 2.1; switch the Model on older nodes to keep them running.

Up to 10 images per request. Each can act as an edit target, subject, object or style reference, so tell the prompt what each one is for, for example 'the person from the first image wearing the coat from the second'. Clear, well-lit shots, ideally a multi-angle character sheet, hold identity best.

It is weak at pixel-exact masked inpainting, vector or flat logo art, seed-locked reproducible output and very stylized painterly looks. Small or long text can blur at 1K, and identity can drift over long edit chains. The lite and 2.5 tiers also don't guarantee true 4K, since they render at the nearest size they support.

Google DeepMind made it. The family began with Gemini 2.5 Flash Image on 26 August 2025. Nano Banana 2.1, the newest tier and this node's default, was released on 6 October 2026.