Models & Pricing

This guide covers the AI models available in Node Banana and their associated costs.

Multi-Provider Support

Node Banana supports multiple AI providers for image and video generation:

• Gemini — Google's image generation models
• Replicate — Wide selection of community and custom models
• fal.ai — Fast inference for popular models

Configure API keys for each provider in Project Settings → Providers tab.

Image Generation Models

Gemini Models

Node Banana uses Google Gemini models for image generation.

nano-banana

The standard model for everyday use.

nano-banana-pro

The premium model for high-quality output.

Pro Model Resolution Pricing

Google Search Feature

The Pro model can use Google Search to inform generation:

• Enables current, real-world references
• Useful for recognizable subjects
• Toggle in node settings

Replicate & fal.ai Models

Replicate and fal.ai provide access to a wide variety of image generation models:

• Model Discovery — Browse available models via the Model Search dialog
• Dynamic Parameters — Each model exposes its own configuration options
• Image-to-Image — Many models support image input for style transfer and modifications
• Pricing — Varies by model; check provider documentation for current rates
• Recently Used — Quick access to your frequently used models

Use the model selector in Generate Image nodes to browse and search available models from all providers.

Video Generation Models

Video generation is available through compatible models on Replicate and fal.ai providers.

Features

• Extended processing — 10-minute timeout for video generation
• Format support — Automatic handling of various video formats
• Video preview — In-node and Output node video playback
• Carousel history — Browse previous video generations

Pricing Considerations

Video generation typically:

• Takes longer than image generation (minutes vs. seconds)
• Costs more per generation
• Varies significantly by model and video length
• Check specific model pricing on provider websites

Choosing Video Models

When selecting a video model, consider:

• Duration — Longer videos cost more
• Resolution — Higher resolution increases cost and processing time
• FPS — Frame rate affects both quality and cost
• Provider — Compare pricing between Replicate and fal.ai

Text Generation Models

Google Gemini Models

OpenAI Models

Cost Tracking

Viewing Costs

Node Banana tracks costs in several places:

1. Header — Shows total workflow cost in dollar format
2. Cost Dialog — Provider-grouped breakdown with detailed cost information
3. Node preview — Shows estimated cost before running
4. Generation history — Tracks cost per generation

The Cost Dialog provides:

• Costs organized by provider (Gemini, Replicate, fal.ai)
• Direct links to external provider pricing pages
• Detailed pricing data for Gemini and fal.ai models
• Dollar format for all costs

How Costs Are Calculated

• Before running: Estimated based on model and settings
• After running: Actual cost recorded
• Cumulative: All costs tracked across sessions
• Provider-specific: Different providers may have different pricing structures

Managing Costs

Tips to reduce costs:

1. Use nano-banana for testing — It's ~3x cheaper than Pro
2. Lower resolution — 1K is often sufficient for iteration
3. Batch carefully — Review prompts before running multiple generations
4. Lock groups — Skip unnecessary nodes during partial runs

Choosing the Right Model

For Iteration/Testing

• Model: nano-banana
• Resolution: 1K
• Cost: ~$0.04/image

For Quality Previews

• Model: nano-banana-pro
• Resolution: 2K
• Cost: ~$0.13/image

For Final Output

• Model: nano-banana-pro
• Resolution: 4K
• Google Search: If beneficial
• Cost: ~$0.24/image

Aspect Ratios

Both models support these aspect ratios:

API Quotas

Google Gemini

• Free tier available with generous limits
• Check Google AI pricing (opens in a new tab) for current quotas
• Rate limits apply to requests per minute

Replicate

• Pay-per-use pricing based on compute time
• Check Replicate pricing (opens in a new tab) for current rates
• Costs vary by model hardware requirements

fal.ai

• Fast inference with per-request pricing
• Check fal.ai pricing (opens in a new tab) for current rates
• Optimized for speed and cost efficiency

OpenAI

• Pay-per-use pricing
• Check OpenAI pricing (opens in a new tab) for current rates
• Rate limits based on your plan tier

Example Cost Scenarios

Quick Test Session

• 10 generations with nano-banana
• Cost: ~$0.39

Standard Workflow Run

• 5 nano-banana (iteration)
• 2 nano-banana-pro (finals)
• Cost: ~$0.47

Production Batch

• 20 nano-banana-pro at 2K
• Cost: ~$2.68