Skip to main content

​
Getting Started

  1. Sign up at app.flowscale.ai
  2. Create a project from your dashboard
  3. Import a workflow by uploading a ComfyUI JSON file, or
  4. Start with a template from our pre-built collection
  5. Test the workflow in the cloud editor
  6. Deploy when you’re ready to create an API
See our Import & Create Workflows guide for detailed steps.
FlowScale supports several import formats:
  • Workflow JSON: Standard ComfyUI export format
  • PNG with metadata: Images with embedded workflow data
  • Workflow templates: FlowScale-specific templates
The platform automatically detects the format and converts it appropriately.
If your models aren’t appearing:Check file format:
  • Supported: .safetensors, .ckpt, .pth, .bin
  • Ensure files aren’t corrupted during upload
Verify upload completion:
  • Large models may take time to process
  • Check the upload progress indicator
Model placement:
  • Models appear in the appropriate category (Checkpoints, LoRAs, VAEs)
  • Use the search function if you have many models
Refresh the interface:
  • Sometimes a browser refresh resolves display issues
  1. Go to project settings (gear icon in your project)
  2. Click “Team Members” tab
  3. Click “Invite Member”
  4. Enter email address and select role level
  5. Send invitation - they’ll receive an email to join
Role levels:
  • Admin: Full project management access
  • Owner: Complete control including billing

​
Workflow Issues

Common causes and solutions:Missing models:
  • Check if all required models are uploaded to your project
  • Verify model names match exactly (case-sensitive)
  • Ensure models are in the correct folders (checkpoints/, loras/, etc.)
Model path issues:
  • Use relative paths in your workflow, not absolute paths
  • FlowScale automatically maps model paths to the correct locations
Model compatibility:
  • Ensure model format is supported (.safetensors, .ckpt, .pth)
  • Check if the model is compatible with your ComfyUI nodes
Quick fix: Go to the node causing the error and re-select the model from the dropdown.
Environment differences:Custom nodes:
  • Install required custom nodes in FlowScale
  • Some nodes may not be compatible with cloud execution
Model versions:
  • Ensure you’re using the same model versions
  • Check if models were uploaded correctly
File paths:
  • Remove any hardcoded local file paths
  • Use FlowScale’s file upload nodes instead
Dependencies:
  • Some workflows may require specific Python packages
  • Contact support if you need additional dependencies installed
Performance optimization:GPU selection:
  • Use higher-tier GPUs (L4, L40S, A100, H100, H200) for complex workflows
  • T4 and A10G GPUs are suitable for simple to moderate workflows
  • B200 for cutting-edge models requiring maximum performance
Model optimization:
  • Use FP16 precision when possible
  • Consider smaller model variants for faster processing
Workflow structure:
  • Minimize unnecessary nodes and connections
  • Use batch processing for multiple similar operations
  • Cache results when appropriate
Cold start delays:
  • First execution may be slower due to model loading
  • Subsequent runs are typically much faster
Memory management:Reduce batch size:
  • Lower the batch_size parameter in your nodes
  • Process images individually rather than in batches
Image resolution:
  • Use smaller image dimensions for development
  • Upscale as a separate step if needed
Model selection:
  • Use smaller models or quantized versions
  • Unload models when not needed
GPU upgrade:
  • Switch to a GPU with more VRAM (L4, L40S, A100, H100, H200)
  • Consider the memory requirements of your specific workflow
  • H200 provides maximum VRAM (141GB) for the largest models

​
Deployment Issues

Deployment troubleshooting:Workflow validation:
  • Ensure workflow runs successfully in the editor first
  • Check for any error nodes or warnings
Input/output configuration:
  • Verify all required inputs are properly configured
  • Ensure output nodes are correctly set up
Resource requirements:
  • Check if your workflow requires specific GPU types
  • Verify sufficient resources are available
Network issues:
  • Temporary network problems can cause deployment failures
  • Try deploying again after a few minutes
If issues persist, check the deployment logs or contact support.
API debugging:Input validation:
  • Check API request format matches documentation
  • Verify required parameters are included
  • Ensure data types match expectations (string, number, etc.)
Authentication:
  • Confirm API key is correct and has proper permissions
  • Check if API key has expired or been revoked
Response format:
  • Review API documentation for expected response structure
  • Check if the workflow output nodes are configured correctly
Rate limiting:
  • Ensure you’re not exceeding API rate limits
  • Implement proper retry logic with exponential backoff
UI troubleshooting:Browser compatibility:
  • Use a modern browser (Chrome, Firefox, Safari, Edge)
  • Clear browser cache and cookies
  • Disable browser extensions that might interfere
Network connectivity:
  • Check your internet connection
  • Try accessing from a different network
Input handling:
  • Ensure uploaded files meet size and format requirements
  • Check if all required fields are filled
Performance:
  • Large files may take time to upload and process
  • Be patient with complex workflows

​
Account & Billing

Billing structure:Compute time:
  • Charged per second of GPU usage
  • Different rates for T4, A10G, L4, L40S, A100, H100, H200 and B200 GPUs
  • $0.07 per hour is charged when pods are idle
Storage:
  • Model storage: $0.01 per model download
  • Workflow storage: Included in base plan
API calls:
  • No additional charges for API requests
Plan management:
  1. Go to account settings (profile icon → Settings)
  2. Click “Billing & Usage” tab
  3. Select “Change Plan”
  4. Choose your new plan
  5. Confirm changes - billing adjusts automatically
Available upgrades:
  • Higher GPU quotas and access to premium GPUs
  • Increased storage limits
  • Priority support and SLA guarantees
  • Team collaboration features
Refund policy:Compute credits:
  • Unused compute time doesn’t expire
  • Credits roll over month to month
  • Can be used across all projects in your account
Subscription refunds:
  • Pro-rated refunds available for annual plans
  • Monthly plans: No refunds, but you can downgrade
Service issues:
  • Credits issued for platform outages or service issues
  • Contact support for technical problems that prevent usage
For specific billing questions, contact our support team.

​
Technical Support

Support channels:In-app support:
  • Click the ”?” icon in any FlowScale interface
  • Submit tickets directly from the platform
  • Include relevant project and workflow details
Community:
  • Discord community for peer support
  • GitHub discussions for feature requests
  • Documentation feedback via GitHub issues
Enterprise support:
  • Dedicated support for Pro and Enterprise plans
  • Phone and video call support available
  • Custom SLA agreements
Response times:
  • Free tier: 48-72 hours
  • Pro tier: 24 hours
  • Enterprise: 4-8 hours (or per SLA)
Include these details:Account information:
  • Your FlowScale username/email
  • Project name and ID
  • Workflow name and version
Problem description:
  • What you were trying to do
  • What actually happened vs. expected behavior
  • Error messages (exact text or screenshots)
  • Steps to reproduce the issue
Environment details:
  • Browser and version (for UI issues)
  • API client and version (for API issues)
  • Any relevant workflow or deployment configurations
Additional context:
  • When the issue started
  • Any recent changes you made
  • Whether the issue is reproducible
Bug reports:
  • Use in-app support for bugs affecting your work
  • GitHub issues for general platform bugs
  • Include reproduction steps and expected vs. actual behavior
Feature requests:
  • Discord community for discussion and voting
  • GitHub discussions for detailed proposals
  • In-app feedback for workflow-specific needs
Security issues:
  • Email security@flowscale.ai for security vulnerabilities
  • Include detailed reproduction steps
  • Do not post security issues publicly
Roadmap:
  • Check our public roadmap for planned features
  • Vote on existing requests to show interest
  • Engage with the community for feature discussion

​
Best Practices

Performance optimization tips:Workflow design:
  • Minimize unnecessary nodes and connections
  • Use efficient node types when available
  • Avoid redundant operations
Model management:
  • Use appropriate model sizes for your use case
  • Consider quantized models for faster inference
  • Cache frequently used models
Resource selection:
  • Choose the right GPU type for your workload
  • Monitor resource utilization and adjust as needed
  • Use batch processing for multiple similar requests
Testing approach:
  • Test with smaller parameters during development
  • Use lower resolution for initial testing
  • Scale up gradually to production settings
Security recommendations:API keys:
  • Keep API keys secure and never share them publicly
  • Rotate keys regularly
  • Use different keys for different environments (dev/prod)
Access control:
  • Use least-privilege principle for team member roles
  • Regularly review and update team permissions
  • Remove access for former team members promptly
Data handling:
  • Don’t include sensitive data in workflows
  • Use secure methods for handling user uploads
  • Understand data retention and deletion policies
Network security:
  • Use HTTPS for all API communications
  • Implement proper input validation

​
Still Need Help?

Can’t find the answer you’re looking for? Here’s how to get additional support:

Join Discord Community

Chat with other FlowScale users and get quick help

Contact Support

Submit a support ticket for technical assistance

Documentation

Explore detailed guides and tutorials

API Documentation

Complete API reference and examples
⌘I