Deploy Workflows

Deploy your ComfyUI workflows as production APIs with automatic documentation, interactive interfaces, and enterprise scaling.

Haven’t set up your workflow yet? You’ll need to complete the Import/Create Workflows guide first. The deployment process requires a fully configured, tested workflow as its foundation.

Prerequisites

Ensure your workflow is ready for deployment, you can check which workflows are ready to deploy from the workflows bar. The workflows with green check mark are ready to deploy:

Successful Execution

Run your workflow successfully in ComfyUI workspace to verify functionality

Save Workflow

Save the workflow after successful execution to create a stable version

Verify Dependencies

Verify all models and custom nodes are properly configured

Workflow tab highlighting green checkmark next to workflow name

Important: The workflow must run successfully at least once in the ComfyUI workspace before deployment. This ensures all models and custom nodes are properly configured and the platform can verify the workflow’s functionality.

Configure API Interface

Transform your workflow into a standardized API by defining inputs and outputs:

Create intuitive parameters for users
Generate automatic documentation
Build interactive interfaces
Establish reliable API contracts

By default, workflows are not configured for deployment. You’ll notice a red cross mark next to unconfigured workflows in your project’s workflow sidebar.

Configure which inputs are user-controllable and how outputs should be formatted for different consumer types.

Configuration Methods

There are two primary ways to configure a workflow for deployment:

FlowScale I/O Nodes (Recommended)
Advanced I/O Configurator

Use specialized FlowScale I/O nodes that are designed for deployment configuration.

FlowScale I/O Nodes

Use FlowScale I/O nodes to define API inputs and outputs. Choose which controls users need without exposing unnecessary complexity.

Adding Input Nodes

Open Node Search

Double-click on an empty area in the ComfyUI workspace to open the node search

Search for Input Nodes

Type one of the following input node names: - InputText - For text prompts and string inputs - InputImage - For image uploads - InputNumber - For numeric parameters - InputSlider - For slider controls

Select FlowScale Nodes

Choose nodes with the “[FS]” prefix, indicating they are FlowScale-specific nodes

Connect to Workflow

Connect these input nodes to your workflow where external inputs are needed

ComfyUI node search dialog displaying FlowScale-specific I/O nodes with [FS] prefix, showing input and output node categories with clear descriptions and usage examples

ComfyUI workspace showing FlowScale I/O nodes being added to workflow with connection preview and configuration options visible

Adding Output Nodes

Search for Output Nodes

In the node search, type one of the following output node names: - SaveImage - For image outputs - SaveText - For text outputs - SaveVideo - For video outputs - SaveAudio - For audio outputs

Select FlowScale Nodes

Again, choose nodes with the “[FS]” prefix

Connect to Workflow

Connect these output nodes to capture the final results of your workflow

Available FlowScale I/O Nodes

Node Type	Purpose	Use Case
[FS] InputText	Text input field	Prompts, descriptions, parameters
[FS] InputImage	Image upload	Reference images, masks, source images
[FS] InputNumber	Numeric input	Steps, strength, dimensions
[FS] InputSlider	Range slider	CFG scale, denoise strength
[FS] SaveImage	Image output	Generated images, processed images
[FS] SaveText	Text output	Generated text, metadata
[FS] SaveVideo	Video output	Animated sequences
[FS] SaveAudio	Audio output	Generated audio files

Verify Configuration

After adding FlowScale I/O nodes:

Connect All Nodes

Ensure all I/O nodes are properly connected to your workflow

Save Workflow

Click the Save button in the ComfyUI workspace

Check Status

Look for a green checkmark next to your workflow in the left sidebar, indicating successful configuration

Deploy Workflow

Deploy your configured workflow as a production API with automatic scaling and monitoring.

Deployment Process

Click Deploy

Click the “Deploy” button in the top-right corner of the application

Main application interface with prominently highlighted Deploy button in top-right corner, showing deployment readiness and access controls

Select Pod

A dialog box will appear asking you to select the Pod for deployment

Choose Configuration

Select the appropriate pod based on your workflow requirements: - GPU type needed for your models - Memory requirements for your workflow - Expected usage patterns

Confirm Deployment

Click Deploy to start the deployment process

Real-time deployment progress interface showing infrastructure provisioning, model loading, and service initialization with progress indicators and status messages

For detailed information about pod types, specifications, and choosing the right configuration for your workflow, refer to our Pods documentation.

Deployed Workflow Features

Your deployed workflow provides:

Real-time monitoring and analytics
Interactive API documentation
Live testing environment
Security and access controls
Performance insights and metrics

Comprehensive API management dashboard showing deployed workflows, request metrics, endpoint URLs, and management controls with professional monitoring interface

API Integration

Your workflow is now accessible via REST API from any application or system:

curl -X POST https://api.flowscale.ai/v1/run \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "workflow_id": "wf_abc123",
    "inputs": {
      "prompt": "A beautiful sunset over mountains",
      "steps": 20,
      "cfg": 7.5
    }
  }'

Interactive Playground

The auto-generated playground provides a user-friendly interface for non-technical users:

Access Playground

In the API Server page, find your deployed workflow and click “Open Playground”

Test Workflow

Use the auto-generated form to input parameters and test your workflow

Share with Team

Share the playground URL with team members for easy access without ComfyUI complexity

List view of deployed workflows with Open Playground buttons, status indicators, performance metrics, and management actions clearly visible

Auto-generated Gradio web interface showing user-friendly form inputs based on workflow parameters, real-time preview, and results display area

Benefits of the playground interface:

Accessible to non-technical team members
Faster feedback cycles on actual results
Interactive client demonstrations
Mobile-friendly access
User confidence through direct testing

API Documentation

FlowScale AI automatically generates comprehensive API documentation:

Interactive docs: Test endpoints directly in the browser
Code examples: Ready-to-use snippets in multiple languages
Schema definitions: Detailed input/output specifications
Authentication: API key management and security

Production Considerations

Security & Authentication

API keys: Secure access with regeneratable keys
Rate limiting: Protect against abuse and manage costs
CORS configuration: Control cross-origin access
Webhook support: Real-time notifications for long-running jobs

Performance Optimization

Caching Strategies

Model caching to reduce cold start times - Result caching for identical requests - CDN integration for faster content delivery

Request Optimization

Async processing for long-running workflows - Batch requests to reduce overhead - Progressive loading for large outputs

Resource Management

Auto-scaling based on demand - GPU pooling for efficient utilization - Memory optimization for large models

Next Steps

Your workflow is now deployed as production infrastructure. Continue building:

Integrate Into Applications

Learn API integration for web apps, mobile apps, and automation systems

Master Advanced Features

Explore advanced configuration, scaling controls, and governance options

Optimize Performance

Monitor usage, optimize costs, and ensure peak performance

Overview

Glossary

Troubleshooting

Prerequisites