3 Steps to Connect Nano Banana 2 to ComfyUI – Gemini Image Generation Model Switching Complete Tutorial

Using Gemini image generation models in ComfyUI is a common need for many creators. This article will show you how to integrate Nano Banana 2 into ComfyUI in 3 steps, enabling workflows like text-to-image, image-to-image, and supporting one-click switching between NB2, NB Pro, and the entire NB model series.

Core Value: By the end of this article, you'll know how to configure the Gemini image generation API in ComfyUI, master model switching techniques, and quickly set up your AI image generation workflows.

Key Aspects of Integrating Nano Banana 2 with ComfyUI

Nano Banana 2 (gemini-3.1-flash-image-preview) is Google's latest image generation model, and you can easily integrate it using ComfyUI custom nodes.

Detailed Explanation of Nano Banana 2 ComfyUI Integration

The core of integrating Nano Banana 2 into ComfyUI is: choosing the right plugin + configuring the API correctly. Unlike directly calling Google's official API, using APIYI's (apiiyi.com) unified interface lets you seamlessly use the entire Gemini image generation model series within ComfyUI, without dealing with complex network and authentication issues.

The integration method is essentially identical to Nano Banana Pro; the only difference is the model name. If you've already configured NB Pro, you just need to change the model name to gemini-3.1-flash-image-preview.

Why Integrate Nano Banana 2 via ComfyUI?

Compared to direct API calls or other tools, ComfyUI integration offers unique advantages:

• Visual Workflows: Node-based operations are intuitive (what you see is what you get) and easy to debug.
• Batch Processing Capability: Supports queued batch generation, perfect for high-volume image needs.
• Workflow Reusability: Set up a workflow once and reuse it repeatedly, boosting efficiency.
• Strong Extensibility: Can be combined with other ComfyUI nodes for post-processing, super-resolution, and more.
• Local Execution: Workflows run locally; only API calls traverse the network, ensuring greater data security.

Nano Banana 2 Full Series Model Comparison

Before integrating with ComfyUI, let's first understand the full range of Gemini image generation models offered by APIYI:

Nano Banana 2 Supported Aspect Ratios

Nano Banana 2 supports 14 aspect ratios, the most among the three models. You can select them using the aspect_ratio parameter in the ComfyUI node:

🎯 Selection Recommendation: Nano Banana 2 strikes the best balance between image quality and speed, making it the recommended model for daily creative work in ComfyUI. You can get a unified API interface for the entire model series through APIYI (apiyi.com), making it easy to switch models anytime.

Nano Banana 2 ComfyUI Plugin Options

Currently, there are several ComfyUI plugins that support Gemini image generation. We recommend the following two options:

Option A: ComfyUI-IF_Gemini Plugin (Recommended)

ComfyUI-IF_Gemini is currently the most mature Gemini ComfyUI node. It natively supports base_url configuration, making it ideal for integration via APIYI.

Advantages:

• Natively supports custom base_url; you can directly enter the APIYI interface address.
• Supports various workflows like text-to-image, image-to-image, and multi-turn conversations.
• Actively maintained, quick community feedback response, and continuous feature updates.
• Uses environment variables for API key configuration, avoiding hardcoding for secure and convenient use.
• Supports setting advanced parameters like temperature and aspect_ratio.

GitHub Repository: github.com/if-ai/ComfyUI-IF_Gemini

Who it's for: Users who need to deeply utilize Gemini image generation features, want to customize more parameters, and build complex workflows.

Option B: comfyui-gemini-nodes (OpenAI Compatibility Mode)

Developed by jqy-yo, comfyui-gemini-nodes supports integration via OpenAI compatibility mode, suitable for users already familiar with OpenAI interfaces.

Advantages:

• OpenAI compatible interface, zero learning curve for users with existing OpenAI experience.
• Supports customization of three parameters: base_url, api_key, and model.
• Lightweight implementation with fewer installation dependencies, less prone to environment conflicts.
• Parameters are entered directly in the node interface, offering a 'what you see is what you get' experience.

GitHub Repository: github.com/jqy-yo/comfyui-gemini-nodes

Who it's for: ComfyUI beginners, users who want to quickly get started and test results, and developers with existing OpenAI interface experience.

Nano Banana 2 ComfyUI Plugin Comparison

💡 Technical Recommendation: If you're new to ComfyUI, we suggest starting with Option B for a quick setup. If you need more extensive features, Option A is recommended. Both options can access the full range of Gemini models via APIYI's (apiyi.com) unified interface.

Getting Started with Nano Banana 2 ComfyUI

Setup Option A: ComfyUI-IF_Gemini

Step 1: Install the Plugin

# Navigate to ComfyUI custom nodes directory
cd ComfyUI/custom_nodes

# Clone the plugin repository
git clone https://github.com/if-ai/ComfyUI-IF_Gemini.git

# Install dependencies
cd ComfyUI-IF_Gemini
pip install -r requirements.txt

Step 2: Configure API (Crucial Step)

Create or edit the .env file in your ComfyUI root directory:

# APIYI Unified API Configuration
GEMINI_BASE_URL=https://vip.apiyi.com/v1
GEMINI_API_KEY=your_apiyi_api_key_here

🚀 Quick Start: Register at APIYI apiyi.com to get your API key. It supports unified model invocation for the entire Gemini image generation model series.

Step 3: Select a Model to Generate Images

After restarting ComfyUI, find the IF_Gemini node in the node panel and set the model parameter:

model: gemini-3.1-flash-image-preview

To switch to other models, just change this one parameter:

# Nano Banana 2 - Pro-grade quality, fastest speed
model: gemini-3.1-flash-image-preview

# Nano Banana Pro - Highest quality
model: gemini-3-pro-image-preview

# Nano Banana - Cost-effective choice
model: gemini-2.5-flash-image

Setup Option B: comfyui-gemini-nodes

Step 1: Install the Plugin

cd ComfyUI/custom_nodes
git clone https://github.com/jqy-yo/comfyui-gemini-nodes.git
cd comfyui-gemini-nodes
pip install -r requirements.txt

Step 2: Configure API in the Node

Restart ComfyUI, then add a Gemini node and configure its properties:

base_url: https://vip.apiyi.com/v1
api_key: your_apiyi_api_key_here
model: gemini-3.1-flash-image-preview

Step 3: Run the Workflow

Connect the prompt text input node, then click 'Queue Prompt' to generate an image.

Nano Banana 2 ComfyUI Full Code Example

Minimal Python Invocation Example

If you want to verify API connectivity outside of ComfyUI first, you can use the following code to test:

import openai

client = openai.OpenAI(
    api_key="your_apiyi_api_key",
    base_url="https://vip.apiyi.com/v1"  # APIYI Unified API
)

response = client.chat.completions.create(
    model="gemini-3.1-flash-image-preview",
    messages=[
        {
            "role": "user",
            "content": "Generate a beautiful sunset landscape painting"
        }
    ]
)

print(response.choices[0].message.content)

{
  "nodes": [
    {
      "type": "IF_GeminiGenerateImage",
      "properties": {
        "model": "gemini-3.1-flash-image-preview",
        "prompt": "A beautiful sunset over mountains, oil painting style",
        "aspect_ratio": "16:9",
        "temperature": 0.8
      }
    }
  ],
  "env": {
    "GEMINI_BASE_URL": "https://vip.apiyi.com/v1",
    "GEMINI_API_KEY": "your_apiyi_api_key"
  }
}

Tip: Get free test credits from APIYI apiyi.com to verify API connectivity before building complex workflows.

Nano Banana 2 ComfyUI Workflow Types

Text-to-Image Workflow

This is the most basic and commonly used workflow: you input a text description to generate an image.

[文本 Prompt] → [Gemini 节点] → [图片预览/保存]

Parameter Suggestions:

• model: gemini-3.1-flash-image-preview
• Aspect Ratio: Choose one of 14 options based on your needs (1:1, 16:9, 9:16, 4:3, 3:4, etc.).
• Temperature: 0.7-0.9 (higher for more creativity, lower for more stability).

Prompt Writing Tips:

• Using English prompts usually yields better results.
• Suggested description order: Subject → Environment → Style → Lighting → Details.
• Example: "A cute cat sitting on a windowsill, soft morning light, watercolor style, detailed fur texture"

Image-to-Image Editing Workflow

Upload an image and use text instructions to edit and modify it:

[加载图片] + [编辑 Prompt] → [Gemini 节点] → [编辑后图片]

Typical Editing Instructions:

• "Change the background to a starry night sky."
• "Add a hat to the person."
• "Convert the photo to a watercolor style."
• "Remove the text watermark from the image."
• "Change the daytime scene to dusk."

Key Usage Points: Nano Banana 2's image-to-image capabilities are very powerful; it can understand image content and precisely execute editing instructions. We recommend making your editing instructions as specific as possible, avoiding vague descriptions.

Style Transfer Workflow

Apply the style of one image to another to create a unique combination:

[风格参考图] + [内容图] + [Prompt] → [Gemini 节点] → [风格化图片]

Common Style Transfer Scenarios:

• Convert photos into artistic styles like oil painting, watercolor, or sketch.
• Transform product images into a specific brand's visual style.
• Convert ordinary photos into anime or illustration styles.

Multi-turn Workflow

Nano Banana 2 supports multi-turn conversational image generation and editing, a unique capability not found in traditional Stable Diffusion models:

第1轮: [Prompt: 画一只猫] → [生成猫图片]
第2轮: [上一轮图片 + Prompt: 给猫戴上红色围巾] → [编辑后图片]
第3轮: [上一轮图片 + Prompt: 把背景换成雪景] → [最终图片]

Advantages: Refining images through multiple iterations offers more control than a single, complex prompt.

Nano Banana 2 ComfyUI Recommended Workflow Configurations

💰 Cost Optimization: For daily creative work, we recommend using Nano Banana 2. Invoking it via APIYI apiyi.com costs just $0.03/invocation, balancing image quality and cost. For batch generation scenarios, you can switch to Nano Banana ($0.025/invocation) to further reduce costs.

Nano Banana 2 ComfyUI Model Switching Tips

Core Method for One-Click Switching

The entire model series is invoked via a unified APIYI apiyi.com interface. To switch models, you just need to change one parameter:

# Method: Only change the model parameter, others remain the same

# Nano Banana 2 - Daily preferred choice
model = "gemini-3.1-flash-image-preview"

# Nano Banana Pro - For ultimate image quality
model = "gemini-3-pro-image-preview"

# Nano Banana - For cost-effectiveness
model = "gemini-2.5-flash-image"

In the ComfyUI node, simply switch the corresponding model name in the model dropdown or text box. The base_url and API Key do not need to be changed at all.

Model Switching Decision Flow

Choose a model based on your creative needs:

需要最高画质? → NB Pro (gemini-3-pro-image-preview)
     ↓ 否
需要 4K + 高速? → NB2 (gemini-3.1-flash-image-preview)
     ↓ 否
追求最低成本? → NB (gemini-2.5-flash-image)

Other Nano Banana 2 ComfyUI Plugins

Besides the two recommended solutions, there are other related plugins in the community you might want to check out:

Note: ComfyUI's official Partner Node doesn't support third-party API integration. If you need to use it via APIYI, please choose Solution A or Solution B. The base_url support for other community plugins varies by version, so please check the documentation in their respective GitHub repositories before use.

Nano Banana 2 ComfyUI Plugin Installation Notes

When installing ComfyUI custom nodes, please keep the following common issues in mind:

1. Python Environment: Make sure to install dependencies using ComfyUI's built-in Python environment to avoid mixing system Python and ComfyUI Python.
2. Dependency Conflicts: If you install multiple Gemini-related plugins, you might encounter dependency version conflicts. It's recommended to only keep one.
3. Plugin Updates: Regularly update the plugin code via git pull to get the latest features and fixes.
4. Log Troubleshooting: When you encounter issues, check the ComfyUI console output; it usually provides clear error messages.

# Update the plugin to the latest version
cd ComfyUI/custom_nodes/ComfyUI-IF_Gemini
git pull
pip install -r requirements.txt

Frequently Asked Questions

Nano Banana 2 ComfyUI Integration Summary

Integrating Nano Banana 2 with ComfyUI is super straightforward, with core steps being: Install Plugin → Configure API → Select Model.

Quick Overview of Key Steps

Key Conclusions

1. Choose the Right Plugin: We recommend ComfyUI-IF_Gemini (feature-rich) or comfyui-gemini-nodes (quick to get started).
2. Configure APIYI's Unified Interface: Set your base_url to https://vip.apiyi.com/v1. Configure it once, and you're ready to use all models in the series.
3. Flexible Model Switching: Easily switch between NB2 ($0.03), NB Pro ($0.05), and NB ($0.025) by just modifying the model parameter.
4. Multiple Workflows: Supports text-to-image, image-to-image, style transfer, and multi-turn conversation to meet various creative needs.
5. Exclusive Feature: NB2's unique Image Search Grounding feature brings even more possibilities to your creations.

We recommend getting your API key quickly through APIYI at apiyi.com. In just 5 minutes, you can experience Nano Banana 2's Pro-level image quality and lightning-fast generation in ComfyUI. The entire model series uses a unified interface, allowing you to switch as needed, anytime.

Author: APIYI Team | Technical Exchange: APIYI apiyi.com

This article is based on actual testing with the latest ComfyUI version and the APIYI platform. For any plugin updates, please refer to the documentation in the respective GitHub repository.