home / mcp / nan banana mcp server
Provides AI-powered image generation via Gemini models with automatic model selection and flexible output controls.
Configuration
View docs{
"mcpServers": {
"zhongweili-nanobanana-mcp-server": {
"command": "uvx",
"args": [
"nanobanana-mcp-server@latest"
],
"env": {
"GCP_REGION": "us-central1",
"GCP_PROJECT_ID": "my-gcp-project",
"GEMINI_API_KEY": "your-gemini-api-key-here",
"IMAGE_OUTPUT_DIR": "/path/to/images",
"NANOBANANA_AUTH_METHOD": "vertex_ai"
}
}
}
}You can run and use the Nano Banana MCP Server to access AI-powered image generation through Gemini models with smart model selection. This server exposes multiple Gemini models, handles authentication, and provides convenient options for deployment, configuration, and image output control to fit production or development needs.
You will connect to the Nano Banana MCP Server using an MCP client. Start the server with one of the available runtime commands, then point your MCP client to the local or remote server configuration. The server automatically selects the best Gemini model based on your prompt or explicit model_tier settings, and it supports grounding, multi-object scenes, aspect ratio control, and output file management.
Prerequisites you need before installation are a Gemini API key and a Python 3.11+ environment for development and testing.
# Prerequisites
- Gemini API key
- Python 3.11+
# Install using the MCP registry (recommended)
uvx nanobanana-mcp-server@latest
# Or install with pip
pip install nanobanana-mcp-serverConfiguration and runtime details are provided to help you securely connect, authenticate, and select models. You can run multiple MCP server configurations to fit your environment, including production deployments on Google Cloud with Vertex AI ADC or local development using an API key.