gpu catalog: a100 40gb
The original A100: same compute as the 80 GB part with half the memory, often the cheapest NVLink-capable data center silicon on the market.
| GPU | VRAM | Regions | Availability | Price / GPU | |
|---|---|---|---|---|---|
| A100 40GBcommunity | 40G | 2 regions | running low | $0.75/hr | launch |
The original A100: same compute as the 80 GB part with half the memory, often the cheapest NVLink-capable data center silicon on the market. VRAM40 GB HBM2 Memory bandwidth1.6 TB/s FP16 tensor312 TFLOPS ArchitectureAmpere InterconnectNVLink 3 (600 GB/s) Board power400 W CUDA cores6,912 Released2020 Regionseu-west, us-east | |||||
| RTX PRO 6000 | 96G | 3 regions | high availability | $1.90/hr | launch |
| RTX A6000 | 48G | 2 regions | high availability | $0.51/hr | launch |
| L40 | 48G | 1 region | high availability | $0.88/hr | launch |
| L40S | 48G | 2 regions | high availability | $0.90/hr | launch |
| RTX 5090community | 32G | 8 regions | high availability | $0.66/hr | launch |
| RTX 4090community | 24G | 7 regions | high availability | $0.38/hr | launch |
| RTX 3090community | 24G | 5 regions | high availability | $0.18/hr | launch |
| H200 SXMcommunity | 141G | 5 regions | available | $4.04/hr | launch |
| H200 SXM | 141G | 5 regions | available | $5.95/hr | launch |
| RTX 5080community | 16G | 5 regions | available | $0.23/hr | launch |
| H100 SXM | 80G | 4 regions | available | $4.94/hr | launch |
| A100 | 80G | 3 regions | available | $1.41/hr | launch |
| RTX 3080community | 10G | 4 regions | available | $0.14/hr | launch |
| A16 | 16G | 6 regions | available | $0.53/hr | launch |
| RTX A4000community | 16G | 3 regions | available | $0.11/hr | launch |
| H200 NVLcommunity | 141G | 5 regions | available | $4.06/hr | launch |
| RTX 4080community | 16G | 6 regions | available | $0.26/hr | launch |
| A40 | 48G | 3 regions | running low | $0.66/hr | launch |
| RTX 6000 ADA | 48G | 1 region | running low | $0.73/hr | launch |
| L40Scommunity | 48G | 2 regions | running low | $0.90/hr | launch |
| L4community | 24G | 2 regions | running low | $0.50/hr | launch |
| RTX 2000 ADA | 16G | 1 region | running low | $0.27/hr | launch |
| RTX PRO 4500 | 32G | 1 region | running low | $0.81/hr | launch |
| RTX 4090 | 24G | 1 region | running low | $1.00/hr | launch |
| RTX 5090 | 32G | 1 region | running low | $1.56/hr | launch |
| B200 | 192G | 1 region | running low | $8.99/hr | launch |
| B200community | 192G | 3 regions | running low | $11.24/hr | launch |
| H100 SXMcommunity | 80G | 1 region | running low | $4.17/hr | launch |
| V100community | 16G | 1 region | running low | $0.15/hr | launch |
| A100community | 80G | 1 region | running low | $1.20/hr | launch |
| RTX A4000 | 16G | 1 region | running low | $0.13/hr | launch |
| H200 NVL | 141G | 1 region | running low | $3.70/hr | launch |
lowest live on-demand rate per GPU, refreshed continuously · per-second billing
Our floor price for the A100 40GB, sampled continuously across every provider in the fleet and rolled up to one point per day. A dot on the zero line is a day we checked and found nothing launchable; a shaded break is a day we did not sample. Those are different facts, so this chart never draws one as the other.
Daily trend, updated hourly · Full 15-minute history available via API
| Provider | Price / GPU / hr | Notes |
|---|---|---|
| GPU.ai (Secure Cloud) | — | Vetted data center capacity, per-second billing |
| GPU.ai (Community Cloud) | $0.750/hr | Marketplace capacity from independent suppliers |
competitor rates are public on-demand list prices, refreshed daily · gpu.ai prices are live floor rates
faq
A100 40GB cloud instances on GPU.ai currently start at $0.750 per GPU per hour, billed per second. That is the live floor price across every provider in our fleet. We route your launch to the cheapest provider with stock.
The NVIDIA A100 40GB has 40 GB of HBM2 memory with up to 1,555 GB/s of bandwidth. GPU.ai reports the minimum VRAM actually delivered across our fleet, so the figure you see at launch is the figure you get.
GPU.ai aggregates a dozen-plus GPU clouds and routes each launch to the cheapest one with stock, so the same silicon is typically far cheaper than hyperscaler list rates. The comparison table on this page shows the current public on-demand list prices side by side.
Billing is per second, starting when your instance is SSH-ready and stopping the moment you terminate. There are no minimum commitments, no reservation fees, and no egress surprises: you pay the listed hourly rate pro-rated to the second.