TensorDock aggregates spare capacity from global GPU providers — H100, A100, RTX 4090, RTX A6000 at the lowest price on the market. No vendor lock-in.
Based on 16 user reviews and our independent benchmarks, TensorDock scored 4.4 / 5. A solid choice that delivers on its promises.
Live pricing from TensorDock. Click any plan to view full details on the provider's site.
| Plan | CPU | RAM | Storage | Bandwidth | Location | Price | |
|---|---|---|---|---|---|---|---|
|
RTX 4090
Ubuntu 22.04 with CUDA, PyTorch, TensorFlow · Bare Metal / KVM
|
8 vCPU | 32 GB | 200 GB NVMe | 1 Tbps uplink | 🇺🇸 United States | $0.62/mo |
Get it |
|
A100 40GB
Ubuntu 22.04 with CUDA, PyTorch, TensorFlow · Bare Metal / KVM
|
16 vCPU | 64 GB | 500 GB NVMe | 1 Tbps uplink | 🇺🇸 United States | $1.16/mo |
Get it |
|
H100 80GB
Ubuntu 22.04 with CUDA, PyTorch, TensorFlow · Bare Metal / KVM
|
32 vCPU | 128 GB | 1000 GB NVMe | 1 Tbps uplink | 🇺🇸 United States | $2.69/mo |
Get it |
TensorDock operates across 6 countries to keep your workload close to your users.
Switched from a bigger provider and save about 40% with no performance loss.
Uptime has been flawless. Cannot recommend enough for production workloads.
Had one hiccup in six months, resolved in under an hour. Otherwise perfect.
Great price-to-performance ratio. Migration was smooth and the dashboard is clean.
Great price-to-performance ratio. Migration was smooth and the dashboard is clean.
Network performance is excellent, especially to Europe. Highly recommended.
Switched from a bigger provider and save about 40% with no performance loss.
Great price-to-performance ratio. Migration was smooth and the dashboard is clean.
Been using them for months with zero downtime. Support responds quickly when you need them.
Switched from a bigger provider and save about 40% with no performance loss.
Support is really knowledgeable — actual engineers, not script readers.
Uptime has been flawless. Cannot recommend enough for production workloads.
Switched from a bigger provider and save about 40% with no performance loss.
Switched from a bigger provider and save about 40% with no performance loss.
Docs are excellent and the API is well-documented. Happy customer.
Support is really knowledgeable — actual engineers, not script readers.
Help other developers by sharing your real experience with TensorDock.
Not sold on TensorDock? Here are three strong alternatives worth comparing.
P2P GPU marketplace connecting renters with hosts for 50%+ cheaper compute.
US GPU cloud specializing in affordable A100, H100, and RTX series rentals.
Darling of serious ML engineers — reserved H100 clusters at unbeatable rates.