RentServersRent GPU servers and clusters
NVIDIA H100, H200, B200 and GB200 for training and inference, plus A100 and L40S for budget and media work. On-demand nodes or reserved clusters, sourced from providers with real availability.
Choose a GPU
Median reserved price per GPU-hour shown on each card.
NVIDIA H100
$3.34The H100 is still the default GPU for training and serving large language models. It has the deepest supply of any current-generation NVIDIA part, whi…
80 GB HBM3 (SXM)
NVIDIA H200
$3.84The H200 uses the same Hopper compute as the H100 but with 141 GB of faster HBM3e memory. For memory-bound work, mainly large-model inference and long…
141 GB HBM3e
NVIDIA B200
$5.62The B200 is NVIDIA's Blackwell-generation GPU for 8-GPU HGX servers. It adds FP4 support and roughly doubles Hopper's dense FP8 throughput, with 180 G…
180 GB HBM3e
NVIDIA GB200 NVL72
QuotedThe GB200 NVL72 is a full liquid-cooled rack: 72 Blackwell GPUs and 36 Grace CPUs joined in a single NVLink domain. It isn't sold by the hour like a s…
~186 GB HBM3e
NVIDIA A100
$1.60The A100 is two generations old but still widely available and cheap. For fine-tuning, batch inference and research that doesn't need FP8, it's often …
40 GB or 80 GB HBM2e
NVIDIA L40S
$1.09The L40S is a PCIe card with 48 GB of GDDR6, FP8 support and no NVLink. It's built for inference, image and video generation, and rendering, where one…
48 GB GDDR6 (ECC)
GPU rental prices
| GPU | Memory | Median reserved | Median on-demand | Lowest in-stock |
|---|---|---|---|---|
| NVIDIA H100 | 80 GB HBM3 (SXM) | $3.34 | $3.74 | $1.78 |
| NVIDIA H200 | 141 GB HBM3e | $3.84 | $4.94 | $3.85 |
| NVIDIA B200 | 180 GB HBM3e | $5.62 | $7.99 | $3.75 |
| NVIDIA GB200 NVL72 | ~186 GB HBM3e | Rack-scale, quoted per contract | ||
| NVIDIA A100 | 40 GB or 80 GB HBM2e | $1.60 | $1.79 | $1.09 |
| NVIDIA L40S | 48 GB GDDR6 (ECC) | $1.09 | $1.50 | $0.55 |
Per GPU-hour, USD. Source: GetDeploying GPU price tracker, captured October 16, 2026.
Size it, then get quotes
Use the configurator to estimate a monthly budget at market medians. Providers' quotes for large reservations usually come in lower.