ComfyUI Crash & Error Guides
Step-by-step terminal fixes, package compilation commands, and VRAM optimization techniques for the most common ComfyUI errors.
OOMtorch.cuda.OutOfMemoryError: CUDA out of memory
How to Fix "torch.cuda.OutOfMemoryError" in ComfyUI
Step-by-step solutions to resolve GPU memory allocation crashes in Stable Diffusion, Flux, and SDXL pipelines.
DependencyModuleNotFoundError: No module named 'insightface'
How to Fix "No module named insightface" in ComfyUI
Install precompiled binary C++ wheels for ReActor, InstantID, and Face Analysis nodes without Visual Studio build tools.
DependencyModuleNotFoundError: No module named 'onnxruntime'
How to Fix "No module named onnxruntime" in ComfyUI
Configure ONNX Runtime GPU support with correct CUDA and cuDNN versions.
ModelValueError: Expected 4 channels in latent, got 16
How to Fix "Expected 4 channels in latent, got 16" in ComfyUI
Resolving latent space dimension conflicts between SD 1.5, SDXL, and FLUX.1 models.
SystemAssertionError: Torch not compiled with CUDA enabled
How to Fix "Torch not compiled with CUDA enabled" in ComfyUI
Replace CPU-only PyTorch with a GPU-accelerated CUDA build.
SystemPrompt outputs failed validation: CLIP or VAE required input is missing
How to Fix "Prompt outputs failed validation" in ComfyUI
Troubleshoot missing model connections, disconnected node sockets, or unselected checkpoints.