Skip to main content
ModelPilot

Check your workflow before you rent a GPU.

Find missing custom nodes and model files, then review hardware guidance. Paste a ComfyUI workflow, upload its JSON file, or try the example.

Free · No account needed · No workflow is executed

Export from ComfyUI: Menu → Save (API Format)

Direct answer

What should I do before deploying a custom ComfyUI workflow?

Run the workflow through the free ModelPilot compatibility checker. It reads the ComfyUI JSON, identifies custom node packages, detects model files, estimates VRAM needs, and helps you review dependencies before choosing hardware. It does not execute the workflow or prove that it runs.

Key facts

  • iNo signup is required for the compatibility check.
  • iThe checker supports frontend and API-format ComfyUI workflow JSON.
  • iThe report is most useful before choosing a GPU or trying to deploy on RunPod.

Next step

  1. 1Paste the workflow JSON and review missing nodes, model files, and GPU requirements.
  2. 2Review the report, choose a supported template or custom setup, then check the current quote before starting.

Not a fit when

  • Do not deploy private or licensed model assets until the user has reviewed model and LoRA usage rights.
  • Browse the template pages instead when the user does not have a custom workflow JSON.

What This Tool Checks

Custom Node Resolution

Looks for matching custom-node packages and links the sources it finds. Unresolved nodes stay visible in the report for review.

GPU Requirements

Estimates memory needs from the detected models and nodes. Resolution, frame count and precision also matter. Confirm hardware and a current quote in the deployment review.

Model Detection

Looks for model-file references in supported nodes, including checkpoints, UNETs, VAEs, LoRAs, ControlNets and CLIP models. Custom or dynamically loaded dependencies can need manual review.

Before you start

Review unresolved dependencies and the selected runtime. Finding a file name does not prove the file is installed, licensed or compatible with the selected node version.

How It Works

  1. 1.Export your workflow from ComfyUI using Menu → Save (API Format). This creates a JSON file with all node types and connections.
  2. 2.Paste the JSON above or drag-and-drop the .json file. The workflow JSON is sent securely for server-side analysis and is not stored. No account is needed.
  3. 3.Get a compatibility report: which custom nodes are resolved, which models are referenced and estimated memory needs. Review current hardware prices before starting. A clean report still needs a runtime check before paid execution.

Common ComfyUI Workflow Issues

Missing custom nodes (red boxes)

When you load a workflow and see red “missing node” boxes, it means the required custom node packages aren't installed. This tool identifies which packages you need and where to find them on GitHub.

CUDA out of memory (OOM) errors

Workflows with large models (Wan 2.2 14B, HunyuanVideo, SUPIR upscaling) need more VRAM than consumer GPUs provide. This tool estimates VRAM requirements so you can choose the right GPU before deploying.

Deploying to cloud GPUs

Running ComfyUI workflows on cloud GPUs (RunPod, vast.ai) requires knowing which Docker image, custom nodes, and models to install. This tool helps plan those dependencies. It does not launch a worker or prove that the complete workflow runs.

Supported Workflow Types

Recognizes many standard ComfyUI workflows in frontend or API format, including:

  • Flux Dev / Schnell image generation
  • SDXL and SD 1.5 workflows
  • Wan 2.2 text-to-video and image-to-video
  • HunyuanVideo and LTX 2.3 video
  • SUPIR and SeedVR2 upscaling
  • ControlNet, IP-Adapter, InstantID
  • Qwen Image Edit and generation
  • Custom LoRA and multi-model pipelines

Compatibility depends on exact model files, custom-node versions, resolution, and batch size. A successful check is not an execution guarantee.