Nanobanana

A lightweight CLI tool for generating and editing images using Google's Gemini API.

Best used with your coding agent CLI of choice! This tool pairs excellently with Claude Code and similar AI coding assistants for automated image generation workflows.

Features

• Text-to-image generation - Create images from text prompts
• Image editing - Transform existing images with text instructions
• Multi-image composition - Combine multiple input images
• Model selection - All current Nano Banana models (Lite, 2, Pro) via -model
• Flexible output - Up to 14 aspect ratios and 4 size options (model-dependent)

Requirements

• Go 1.25+ (for building from source)
• Google Gemini API key with access to the Gemini image models (Nano Banana)

Installation

Homebrew (macOS/Linux)

brew tap skorfmann/nanobanana
brew install nanobanana

To upgrade to the latest version:

brew upgrade nanobanana

From GitHub Releases

Download the latest binary for your platform from the Releases page.

# Example for macOS arm64
curl -L -o nanobanana https://github.com/skorfmann/nanobanana/releases/latest/download/nanobanana-darwin-arm64
chmod +x nanobanana
./nanobanana -version

From source

git clone https://github.com/skorfmann/nanobanana.git
cd nanobanana
go build -o nanobanana main.go

Setup

1. Get a Gemini API key from Google AI Studio
2. Set it as an environment variable:

export GEMINI_API_KEY="your-api-key-here"

Usage

nanobanana [options] "prompt"

Options

Models

Full Gemini model IDs (e.g. gemini-2.5-flash-image) are also accepted, so new models work without a rebuild.

Supported Aspect Ratios

All models: 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9

flash only: 1:4, 4:1, 1:8, 8:1 (extreme banners/skyscrapers)

Supported Sizes

Supported Image Formats

PNG, JPEG, WebP, GIF

Examples

Text-to-image

# Simple generation
nanobanana "a cute cat sitting on a windowsill"

# With aspect ratio and size
nanobanana -aspect 16:9 -size 2K "cinematic mountain landscape at sunset"

# Custom output filename
nanobanana -o hero-image.png "abstract geometric pattern"

# Cheap draft with the lite model
nanobanana -model lite "quick concept sketch of a logo"

# Maximum quality with the pro model
nanobanana -model pro -size 4K "detailed product shot of a watch"

# Extreme banner ratio (flash only)
nanobanana -aspect 4:1 -size 0.5K "website header, abstract gradient"

Image editing

# Style transfer
nanobanana -i photo.jpg "transform into watercolor painting"

# Modifications
nanobanana -i portrait.png "add sunglasses"

Multi-image composition

# Combine images
nanobanana -i background.png -i subject.png "place the subject in the scene"

# Style reference
nanobanana -i content.jpg -i style.jpg "apply the style to the content image"

Examples

The examples/ folder contains working examples with generated images:

basic/

Simple text-to-image generation.

./nanobanana -o basic_example.png "a friendly yellow banana character"

presentation/

Generate presentation slides from text prompts.

./nanobanana -aspect 16:9 -size 2K -o slide_01.png "title slide prompt..."
./nanobanana -aspect 16:9 -size 2K -o slide_02.png "content slide prompt..."

branded-presentation/

Use a template image as a style reference for consistent branding across slides.

# 1. Generate a style template first
./nanobanana -aspect 16:9 -size 2K -o template.png "slide template with brand colors..."

# 2. Generate slides using template as reference
./nanobanana -i template.png -aspect 16:9 -size 2K -o slide_01.png "title slide..."
./nanobanana -i template.png -aspect 16:9 -size 2K -o slide_02.png "content slide..."

Each example includes a README and the markdown source used to generate the images. See the examples/ folder for full prompts and generated outputs.

Using with Coding Agents

Nanobanana works great with AI coding assistants like Claude Code for automated image generation workflows:

1. Describe slides/images in a markdown file
2. Your coding agent reads the markdown and extracts prompts
3. The agent runs nanobanana to generate each image

See examples/branded-presentation/ for a complete workflow demonstration.

API Pricing

Approximate cost per image (paid tier, July 2026):

See Gemini API Pricing for current rates.

License

MIT