Boots in minutes
A full KVM virtual machine with the GPU passed through: your kernel, root, systemd.
Template · Base & frameworks
Run NVIDIA NGC PyTorch on a cloud GPU, in minutes
NVIDIA's NGC-optimised PyTorch build — tuned kernels for the datacenter cards. NVIDIA's NGC PyTorch container is the tuned build: pre-compiled kernels for Hopper and Ampere, APEX, DALI, Transformer Engine and FP8 support already wired. It is what large-scale training teams run on datacenter cards; rent it here with the same image, no registry login required. From $0.291/hr on an interruptible A100 SXM4.
Template is free
You pay the GPU price only: PyTorch NGC on a A100 SXM4 is $0.583/hr on-demand, $0.291/hr interruptible, billed per second.
Volumes for state
Models, datasets and outputs live on a $0.08/GB/mo volume; the instance stays disposable.
Private by default
Dedicated GPU, encrypted disk, crypto payments, no KYC, no stored IPs — SSH behind your own credentials.
Best GPUs for PyTorch NGC
Three price points that run this template well — Good, Better, Best. Every model in the catalogue works; these are the value picks.
| Tier | GPU | VRAM | On-demand | Interruptible | Why this card | Action |
|---|---|---|---|---|---|---|
| Good | A100 SXM4 | 80 GB | $0.583 | $0.291 | Ampere datacenter card the NGC stack is tuned for. | Deploy |
| Better | H100 SXM | 80 GB | $1.587 | $0.793 | Transformer Engine FP8 paths shine on Hopper. | Deploy |
| Best | H200 | 141 GB | $3.058 | $1.529 | 141 GB for longer context and bigger micro-batches under the same tuned stack. | Deploy |
Need more VRAM? The full catalogue lists all 76 models with live availability; the VRAM guide sizes models to cards.
Deploy PyTorch NGC from the console, CLI or API
Pick the template in the console deploy bar, or script it:
- Console — filter by GPU, choose PyTorch NGC in the template picker, set disk and env, deploy.
- CLI — pip install powergpu, then the command on the right. CLI reference.
- API — POST /v1/instances with "template": "pytorch-ngc". REST reference.
- Own image — any OCI reference works too; we inject the NVIDIA runtime. Template docs.
$ powergpu launch --gpu a100-sxm4 --template pytorch-ngc \
--disk 100 --volume models:/workspace/models
✓ instance i-7a41c0e2 running (2m38s)
# PyTorch NGC · A100 SXM4 · $0.583/hr · per second
# https://i-7a41c0e2.powergpu.io:8000 (TLS)
$ powergpu stop i-7a41c0e2 # billing ends this secondWhat is inside the PyTorch NGC template
| Image | nvcr.io/nvidia/pytorch |
|---|---|
| CUDA | inherits host driver (any 12.x / 13.x) |
| Access | SSH shell · full virtual machine (KVM) |
| Category | Base & frameworks |
| Storage | Instance NVMe disk (sized at deploy) + optional network volumes |
| Billing | GPU price only, per second — no template fee, no setup fee |
Other base & frameworks templates
NVIDIA CUDA on a cloud GPU — The bare CUDA base image — drivers, toolkit, SSH and Jupyter. Build your own stack on top.
PyTorch on a cloud GPU — The training and research default: PyTorch 2.6, cuDNN, JupyterLab and SSH out of the box.
TensorFlow CUDA on a cloud GPU — TensorFlow with GPU support, Keras and TensorBoard on a mapped port.- NVIDIA RAPIDS on a cloud GPU — GPU-accelerated data science — cuDF, cuML, cuGraph in a notebook.
All-in-One App Studio on a cloud GPU — A launcher bundling the most-used AI apps behind one desktop — pick and run.- All 37 templates
PyTorch NGC on a cloud GPU: FAQ
Environment variables, ports and custom images are covered in the template docs.
How much does it cost to run PyTorch NGC on a cloud GPU?
Only the GPU price — the template is free. From $0.291/hr on an interruptible A100 SXM4, $0.583/hr on-demand on a A100 SXM4. Billing is per second, so an hour of tinkering costs an hour, not a day. Storage is $0.08/GB/month.
How long does PyTorch NGC take to start?
Two to four minutes: this template is a full virtual machine that boots its own kernel. Restarting a stopped instance is faster, and your disk is exactly as you left it.
Can I keep my models and outputs between sessions?
Yes — attach a volume at deploy. Everything on the volume survives instance destruction and mounts on the next instance in the region in seconds, at $0.08/GB/month. The instance disk itself survives stop/start but not destroy.
Why choose the NGC build over plain PyTorch?
On datacenter GPUs the NGC container is typically 10–30% faster on transformer training thanks to Transformer Engine, fused kernels and NCCL tuning — and it is validated release by release by NVIDIA.