ModuleNotFoundError: No module named 'onnx' or 'onnxruntime' in ComfyUI

No module named 'onnx' and No module named 'onnxruntime' appear when an ONNX-backed face, pose, or helper node cannot import its backend. First check whether the failing workflow actually uses ONNX nodes. onnx loads .onnx model files; onnxruntime runs them — they are related but not the same, and not every workflow needs both.

WanVideoWrapper WARNING: FantasyPortrait nodes not available: No module named 'onnx'
No module named 'onnxruntime'
ModuleNotFoundError: No module named 'onnxruntime.capi'
DWPose: Onnxruntime not found or doesn't come with acceleration providers, switch to OpenCV with CPU device

Custom nodes that may need ONNX / ONNX Runtime include ComfyUI-WanVideoWrapper (FantasyPortrait), comfyui_controlnet_aux (DWPose), ComfyUI-ReActor, ComfyUI_IPAdapter_plus, ComfyUI_InstantID, and ComfyUI-PuLID. When missing, those nodes are unavailable while the rest of ComfyUI still works.

Severity: medium — install only if your workflow uses ONNX-backed pose, face, or helper models. A warning from an optional node family you do not use can be ignored.

Triage: pick the right ONNX branch

Stay anchored to the newest error. Do not switch to an older package family just because a previous log line mentioned a different plugin.

Use the correct Python

Plain pip install from a random terminal is the most common reason the error stays. If a tool asks for a package name, enter onnxruntime or onnx, not pip install onnxruntime.

Option 1: Install the CPU version (safest first choice)

python -m pip install onnx onnxruntime

For the official Windows portable package:

.\python_embeded\python.exe -s -m pip install onnx onnxruntime

This is enough for many use cases; features like DWPose work in CPU mode, just slower.

Option 2: Install the GPU version

For an NVIDIA GPU where the workflow benefits from ONNX Runtime GPU acceleration:

python -m pip install onnx onnxruntime-gpu

Do not keep both onnxruntime and onnxruntime-gpu installed. If you had the CPU package, remove it first:

python -m pip uninstall onnxruntime -y
python -m pip install onnxruntime-gpu

Common installation issues

GPU not working after installing onnxruntime-gpu:

[W:onnxruntime:Default, onnxruntime_pybind_state.cc] LoadLibrary failed with error 126 when trying to load onnxruntime_providers_cuda.dll

The CUDA/cuDNN family required by onnxruntime-gpu does not match PyTorch's runtime stack. Use CPU onnxruntime unless the workflow is clearly bottlenecked by ONNX inference; if you need GPU, match ONNX Runtime GPU to your PyTorch CUDA/cuDNN family instead of installing the latest package blindly.

Both onnxruntime and onnxruntime-gpu installed: check with python -m pip list | findstr onnxruntime, then keep only the one you need:

python -m pip uninstall onnxruntime onnxruntime-gpu -y
python -m pip install onnxruntime

protobuf version conflict (TypeError: Descriptors cannot be created directly.) — quote the requirement so shells do not interpret >:

python -m pip install "protobuf>=3.20"

Verify

python -c "import onnxruntime as ort; print(ort.__version__, ort.get_available_providers())"

If GPU acceleration works, CUDAExecutionProvider appears. If only CPUExecutionProvider appears, the CPU package is installed or the GPU provider did not load.

Source References

• ComfyUI troubleshooting overview
• ComfyUI custom node troubleshooting guide
• ComfyUI model troubleshooting guide
• ONNX Runtime installation documentation