Skip to main content
Runners are the compute resources that execute your ComfyUI workflows. ComfyControl supports two types of runners: Self-Hosted Runners that you manage yourself, and Managed Runners that we operate for you.

​
Self-Hosted Runners

Self-hosted runners are ComfyUI instances that you set up and manage. ComfyControl connects to these instances for workflow execution. You have complete control over the infrastructure, models, and custom nodes installed.
You’re responsible for managing AI models and custom nodes on your ComfyUI instances. We do not automatically install dependencies required for workflow execution.
Best for:
  • Organizations with existing ComfyUI infrastructure
  • Users who need full control over their environment
  • Custom hardware configurations
  • On-premises deployments

​
State Management

States only control the scheduling of workflows on runners.
A runner can be in one of two states:
An active runner is one where ComfyControl will schedule workflows. A newly added runner is always in active state.
Inactive runners will never get any workflow for execution.
You can manage the state of runners from their detail page.

​
Tags

Tags help you organize runners and control workflow placement based on hardware capabilities or specialization. You can assign multiple tags to each runner to categorize them by:
  • Hardware specifications (e.g., gpu-4090, high-memory)
  • Specialization (e.g., text-to-image, video-generation)
  • Environment (e.g., production, development)
  • Location or availability zones
Use tags to route specific workflows to runners with the right capabilities. For example, you can ensure video generation workflows only run on high-performance GPU runners.
When creating or scheduling workflows, you can specify required tags to ensure they execute on appropriate runners. This gives you fine-grained control over resource allocation and workflow execution.
Runners without tags act as wildcard runners and can accept workflows regardless of their tag requirements.

​
Authentication

Runners support custom authentication headers to access protected ComfyUI endpoints. This is essential when your ComfyUI instances are behind:
  • API gateways with authentication requirements
  • Reverse proxies requiring authorization tokens
  • Custom security implementations
Configure authentication headers in the runner settings. ComfyControl will include these headers in all requests to the ComfyUI instance.
Common authentication patterns include:
  • Bearer tokens for API authentication
  • Basic authentication credentials
  • Custom header-based security schemes

​
Automatic Health Checks

ComfyControl on demand monitors runner availability and connection status to ensure reliable workflow execution. Health check is only performed when you visit details page which includes:
  • Monitors Connection Status: On demand verifies that runners are accessible and responsive
  • System Stats and Resources: Captures the system stats and varioud version of software such as Python, CUDA etc
Health check and System stats is visible on the runners detail pages, helping you quickly identify and troubleshoot connectivity or performance issues.

​
Managed Runners

Managed runners are high-performance cloud GPUs that we operate and maintain on your behalf. These industry-grade runners feature enterprise GPUs (such as H100, H200, and other powerful hardware) and require zero setup or configuration.
With managed runners, we handle all infrastructure management, model installations, upgrades, and maintenance. You can focus on creating workflows while we ensure optimal performance and availability.

​
Key Benefits

  • Zero Setup: No infrastructure management or ComfyUI installation required
  • Enterprise-Grade Hardware: Access to powerful GPUs like H100, H200, and more
  • Fully Maintained: We handle all upgrades, model management, and custom nodes
  • Run Larger Models Faster: High-performance hardware optimized for demanding workloads
  • Automatic Availability: Instantly available based on your subscription tier

​
Availability & Billing

Managed runners are exclusively available on Managed subscription tiers.
Workflow execution on managed runners is billed per second of GPU usage. Each tier includes a specific allocation of compute minutes per month, allowing you to run workflows without worrying about infrastructure costs.
Check your dashboard to view your current compute minute allocation and usage. Pricing details and tier-specific allocations are available in your account settings or by contacting our sales team.

​
Working with Managed Runners

Viewing Available Runners: You can view all available managed runners in your dashboard. Each managed runner displays its GPU type, specifications, and assigned tags that identify its capabilities. Scheduling Workflows: If your subscription tier includes managed runners, workflows can be automatically scheduled on them. Use the tag matching system (described in the Self-Hosted Runners section above) to target specific GPU types for your workflows. For example, if you need an H100 GPU for a demanding workflow, specify the h100 tag when scheduling. ComfyControl will ensure your workflow runs on the appropriate hardware. Workflow Control: While you cannot configure or manage the runner infrastructure itself (we handle that), you retain full control over your workflows:
  • View workflow status and progress in real-time
  • Cancel any workflow running on a managed runner at any time
  • Monitor execution time and resource usage
What You Don’t Need to Manage: Unlike self-hosted runners, managed runners don’t require:
  • State management (Active/Inactive)
  • Custom authentication configuration
  • Health check monitoring
  • Model or custom node installation
  • Infrastructure upgrades or maintenance
Everything is handled automatically by ComfyControl, allowing you to focus solely on your creative workflows.