torch.cuda.is_available() returns False: causes and fixes (Mac, PC without NVIDIA, Docker)

The one-command diagnosis

Run this script: it tells you which system you are on, whether your PyTorch build targets CUDA, and whether a variable hides the GPU.

import os, platform
import torch

print("system       :", platform.system(), platform.machine())
print("torch        :", torch.__version__)
print("built for CUDA:", torch.version.cuda)  # None = PyTorch without CUDA
print("cuda available:", torch.cuda.is_available())
print("CUDA_VISIBLE_DEVICES:", os.environ.get("CUDA_VISIBLE_DEVICES"))
if torch.cuda.is_available():
    print("GPU:", torch.cuda.get_device_name(0))
elif torch.backends.mps.is_available():
    print("Apple GPU (MPS): no CUDA on this machine")

The causes and their fixes

Reinstall PyTorch with CUDA

If torch.version.cuda is None, you have the CPU build. Reinstall a CUDA build; the selector on pytorch.org gives the exact command for your system and driver.

pip uninstall -y torch torchvision torchaudio
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126
python -c "import torch; print(torch.version.cuda, torch.cuda.is_available())"

In Docker

A container does not see the GPU by default. You must ask for it at startup, and the host needs the NVIDIA driver and the NVIDIA Container Toolkit.

# The container only sees the GPU when you ask for it
docker run --rm --gpus all your-image nvidia-smi

On a Mac: MPS for tests, an NVIDIA GPU for CUDA

Macs have a GPU that PyTorch uses through mps, handy for small tests. But code that requires CUDA (CUDA kernels, bitsandbytes, vLLM, many training scripts) does not run on a Mac. Write your code so it picks its device: CUDA when present, then MPS, then the CPU.

import torch

if torch.cuda.is_available():
    device = torch.device("cuda")
elif torch.backends.mps.is_available():
    device = torch.device("mps")
else:
    device = torch.device("cpu")

model = model.to(device)

The fix without an NVIDIA card: a GPU by the hour

The GPU Cloud PyTorch (CUDA) template delivers an Ubuntu server with the NVIDIA driver, CUDA and PyTorch ready in /opt/pytorch, activated at SSH login. An RTX A6000 with 48 GB costs CA$0.67/h. The guide to run your script on a GPU shows every step, from copying the project to stopping the charges.

Good to know before you launch

• Billing is hourly, at the server’s hourly rate, from prepaid credit. There is no per-minute billing, and the first hour is charged when you deploy.
• A stopped server is still billed at the full hourly rate. To pay less, hibernate it ($0.02 per hour (about $14.60 per month), disk kept) or delete it. A Spot machine cannot be hibernated: delete it when you are done.
• The minimum top-up is CA$25.00, paid by credit card.
• Servers and their disks stay in Canada (region canada-montreal, in Montreal). Account information (identity, billing, e-mails) may be processed outside Quebec, as the privacy policy explains.

Frequently asked questions