Banana Nano: Redefining AI Image Editing & Generation

Experience the next generation of AI image technology powered by Gemini 2.5 Flash Image. Achieve unprecedented image editing, fusion, and creation through simple natural language instructions, turning your imagination into reality.

Revolutionary Editing

Unprecedented image editing and control via natural language.

High Consistency

Maintain character and style consistency across edits and scenarios.

Native Multimodality

Deep semantic understanding of text and images for complex reasoning.

Live Demo

Experience the power of Banana Nano firsthand. Enter a prompt or upload reference images below to generate your own image using the real model.

Reference Images 0/9
Prompt

Ready to Generate

Enter a prompt to unleash the power

Inspiration

Need some inspiration? Click any image below to load its prompt into the demo area above. You can then freely modify and generate your own version.

Character Figure
Turn the photo into a character figure. Behind it, place a box with the character’s image printed on it, and a computer showing the Blender modeling process on its screen. In front of the box, add a round transparent PVC base with the character figure standing on it. Set the scene indoors.

Click to copy prompt

Group Photo Composition
Combine all the people from the photos into a single group photo.

Click to copy prompt

Sketch Character Battle Scene
Have these two characters fight using the pose from Figure 3. Add appropriate visual backgrounds and scene interactions. The generated image ratio is 16:9.

Click to copy prompt

Complex Scene Fusion
On a pure white studio background, a young female model wears a white sailor suit, carries a single-shoulder backpack (with a cat plush toy and marker), has a red ribbon high ponytail, leans against a motorcycle, holds a blue crystal in her left hand and an umbrella in her right. Next to the motorcycle is a can of golden beer and a three-legged chair toy, with an electric violin leaning against the front wheel. The whole scene is bright and exquisite.

Click to copy prompt

1:1 authentic cosplay
1:1 authentic cosplay

Click to copy prompt

Restore and colorize old photos
1:1 authentic cosplay

Click to copy prompt

The Emergence of a Mystery

Its debut came without announcement or launch event, more like a "perfect storm" sweeping through the AI community. A mysterious model quietly appeared in the arena, sparking a global "puzzle game" led by developers and enthusiasts thanks to its undeniable strength. Here are the three key stages of this community-driven legend's unveiling.

1. LMArena Trial

The model quietly appeared on the anonymous AI blind test platform LMArena. Without users knowing its origin, it consistently outperformed well-known models, quickly gaining a "cult-like" following and establishing its technical lead even before its official release.

2. Community Discovery & Speculation

Its mysterious veil and powerful capabilities sparked heated discussions in major tech communities. Through performance traits, naming conventions (Nano + Banana), and hints from Google employees, the AI community gradually deduced it likely originated from Google DeepMind.

3. Official Reveal

On August 26, 2025, Google officially announced that Banana Nano's official name is Gemini 2.5 Flash Image. This "validate first, announce later" strategy successfully leveraged community buzz for an extraordinary pre-launch campaign.

Core Technical Capabilities

Its outstanding performance is rooted in advanced architecture. Click the cards below to learn about its three key pillars.

Consistency & Coherence

Maintains subject identity consistency across multiple edits, scenarios, and even style changes, seamlessly integrating new elements while preserving the original background, lighting, and shadows.

Natural Language Editing

Achieves a new level of understanding for natural language instructions, accurately parsing complex, multi-step prompts. Users can iteratively modify images as if conversing with a designer.

Multi-Image Fusion

Intelligently fuses elements from up to 3 source images into a new, visually harmonious scene, automatically adjusting lighting, perspective, and texture.

Competitive Landscape

How does Banana Nano position itself in a fierce market? Select models below to visually compare their performance on key features using the radar chart.

Frequently Asked Questions (FAQ)

Find answers to your common questions about Banana Nano here.

How is Banana Nano different from other AI image tools?

The main difference lies in its superior "consistency" and "natural language editing" capabilities. It better understands continuous, complex instructions and maintains subject features (such as people or objects) across multiple edits, which is crucial for storytelling and character design.

Do I need professional knowledge to use it?

Not at all. Banana Nano is designed so everyone can create through everyday conversation. You don't need to learn complex "spells" or parameters—just describe your ideas as you would to a designer.

Can the generated images be used commercially?

This usually depends on the terms of service of the platform you use. Images generated via the official API or Google products (such as Vertex AI) follow Google Cloud's relevant policies. Please check the specific platform's terms before use.

How is Banana Nano priced?

Pricing is based on each successful image generation request. As shown in the "Ecosystem" section, the price is highly competitive (about $0.039 per image), aiming to lower the barrier for high-quality AI image creation and make it accessible to more developers and creators.

Ecosystem & Commercialization

With affordable pricing, broad platform coverage, and strategic partnerships, Google is paving the way for large-scale adoption of this technology.

Access Platforms

  • Gemini App
  • Google AI Studio
  • Vertex AI
  • Third-party API

Strategic Partnerships

Adobe

Integrated with Firefly & Express

Pricing Model

~$0.039

/ per image

Safety Technology

SynthID

Visible & invisible digital watermark