Hyperstack vs TensorDock
TensorDock is cheaper on 4 of the 5 GPU configurations both providers list with on-demand pricing, updated 2026-10-02. Hyperstack still wins on one, its best case being the A100 SXM 80GB · node count not reported at $1.60 vs $1.80. Platform facts come from each provider's documentation.
Hyperstack
cheaper on 1 of 5 sharedTensorDock
cheaper on 4 of 5 sharedPrices are per GPU-hour. Rows match exact GPU variants and published node minimums. When node count is not reported, confirm the required configuration before comparing whole-job costs. These are listed rates, not a guarantee of stock.
Price per GPU, side by side (2026-10-02)
| GPU (on-demand $/hr) | Hyperstack | TensorDock | Price diff |
|---|---|---|---|
| RTX A4000 · node count not reported | $0.15 | $0.10 | -$0.05 (-33.3%) TensorDock |
| Tesla V100 SXM2 · node count not reported | not offered | $0.17 | – |
| RTX 3090 · node count not reported | not offered | $0.20 | – |
| RTX 4090 · node count not reported | not offered | $0.35 | – |
| RTX A6000 · node count not reported | $0.50 | $0.45 | -$0.05 (-10.0%) TensorDock |
| RTX 6000 Ada · node count not reported | not offered | $0.75 | – |
| L40 · node count not reported | $1.00 | $0.95 | -$0.05 (-5.0%) TensorDock |
| A100 (unspecified) · node count not reported | $1.35 | not offered | – |
| A100 PCIe 80GB · node count not reported | not offered | $1.50 | – |
| A100 SXM 80GB · node count not reported | $1.60 | $1.80 | -$0.20 (-11.1%) Hyperstack |
| RTX PRO 6000 (unspecified) · node count not reported | $1.85 | not offered | – |
| H100 SXM · node count not reported | $3.20 | $2.25 | -$0.95 (-29.7%) TensorDock |
| H100 (unspecified) · node count not reported | $2.50 | not offered | – |
| H200 (SXM) · node count not reported | $3.99 | not offered | – |
| B200 · node count not reported | $6.00 | not offered | – |
| B300 · node count not reported | $7.40 | not offered | – |
Lowest recorded on-demand rate per configuration; highlighted = cheaper. Price diff is the absolute and percent saving on the cheaper side (named), relative to the higher price; – means one side has no matching configuration in the collected rates. Spot/marketplace tiers shown on the individual pages.
Tables rank by price, lowest first. Commercial relationships never affect ordering.How we collect and rank prices.
Platform facts
| Hyperstack | TensorDock | |
|---|---|---|
| Billing increment | not published | not recorded |
| Egress | not published | not recorded |
| Storage /GB-mo | $0.07/GB/month shared storage | not recorded |
| Spot offered | yes | not recorded |
| Reserved offered | not published | not recorded |
| Notebook support | templates | not recorded |
From each provider's documentation, updated 2026-07-07; sources on the fees matrix and enterprise matrix.
FAQ
TensorDock is cheaper on 4 of the 5 GPU configurations both providers list, updated 2026-10-02. Hyperstack is cheaper on 1; 0 have equal rates. Compare the configuration you need.
The H100 SXM · node count not reported: $2.25 per GPU-hour on TensorDock vs $3.20 on Hyperstack, a difference of $0.95 per hour, updated 2026-10-02.
Hyperstack lists 6 GPU configurations with on-demand pricing that TensorDock does not: A100 (unspecified) · node count not reported, RTX PRO 6000 (unspecified) · node count not reported, H100 (unspecified) · node count not reported, H200 (SXM) · node count not reported, B200 · node count not reported, B300 · node count not reported. TensorDock lists 5 that Hyperstack does not: Tesla V100 SXM2 · node count not reported, RTX 3090 · node count not reported, RTX 4090 · node count not reported, RTX 6000 Ada · node count not reported, A100 PCIe 80GB · node count not reported.
Compare more rental options: the screener · price movers · methodology.