Driver issues
nvidia-smi: command not found
nvidia-smi: command not found
By Fredoline Eruo · Last verified Jun 12, 2026
Cause
Either NVIDIA drivers aren't installed, or nvidia-smi exists but isn't on PATH. On Windows, nvidia-smi.exe lives in C:\Windows\System32\ after driver install. On Linux, it's in /usr/bin/ or /usr/local/cuda/bin/.
Solution
On Linux:
# Find it
which nvidia-smi
locate nvidia-smi
# If found but not on PATH:
export PATH=/usr/local/cuda/bin:$PATH
# If not installed:
sudo apt install nvidia-driver-535 # Ubuntu — pick latest
Then reboot.
On Windows:
# Add to PATH if installed but missing
$env:Path += ";C:\Windows\System32"
# Or run with full path
& "C:\Windows\System32\nvidia-smi.exe"
On WSL2: Don't install NVIDIA drivers inside WSL. Install the Windows driver only — WSL2 inherits it. nvidia-smi should work in your WSL terminal automatically.
If nvidia-smi fails after install: /dev/nvidia0 may not exist. Try sudo nvidia-smi once to initialize, then it works for users.
Related errors
- CUDA driver version is insufficient for CUDA runtime version
- PyTorch CUDA error: driver version is insufficient for CUDA runtime
- WSL2: nvidia-smi works but PyTorch sees no CUDA / libcuda.so missing
- WSL2 GPU not detected — nvidia-smi missing or empty
- Docker container can't see GPU — nvidia-container-toolkit missing
Did this fix it?
If your case was different, email Contact support with what you saw and we'll update the page. If it worked but took different commands on your platform, we want to know that too.