NVIDIA B200 180GB: cloud rental cost
Renting one NVIDIA B200 180GB around the clock costs from $4,365 a month (Runpod Community Cloud, on demand) to $83,171 (AWS EC2, on demand), across 4 providers with a published price. The listed AWS EC2 instance has 8 GPUs, so a single GPU there is billed as 8.
A full eight-GPU setup on demand ranges from $34,923 a month on Runpod Community Cloud to $83,171 on AWS EC2. At the benchmark throughput shown below, the cheapest full-time eight-GPU option works out at $0.145 per million tokens, a best case that assumes the GPUs are busy every hour.
Prices on this page last verified on Sources
1 GPU, 24 hours a day
One GPU serving a model around the clock. Monthly cost, cheapest on-demand provider first.
| Provider | GPUs billed | On demand | Spot | 1-year reserved |
|---|---|---|---|---|
| Runpod Community Cloud | 1 | $4,365$0.145 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| Lambda | 1 | $5,833$0.194 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| Runpod Secure Cloud | 1 | $5,833$0.194 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| AWS EC2 | 8 | $83,171$2.77 per 1M tokens | Price not verified yet. | Price not verified yet. |
8 GPUs, 24 hours a day
A full eight-GPU server running around the clock. Monthly cost, cheapest on-demand provider first.
| Provider | GPUs billed | On demand | Spot | 1-year reserved |
|---|---|---|---|---|
| Runpod Community Cloud | 8 | $34,923$0.145 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| Lambda | 8 | $44,910$0.187 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| Runpod Secure Cloud | 8 | $46,662$0.194 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| AWS EC2 | 8 | $83,171$0.346 per 1M tokens | Price not verified yet. | Price not verified yet. |
1 GPU, 8 hours a day
Development or test work during a working day, switched off overnight. Monthly cost, cheapest on-demand provider first.
| Provider | GPUs billed | On demand | Spot | 1-year reserved |
|---|---|---|---|---|
| Runpod Community Cloud | 1 | $1,455$0.145 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| Lambda | 1 | $1,944$0.194 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| Runpod Secure Cloud | 1 | $1,944$0.194 per 1M tokens | No self-serve price published for this option. | No self-serve price published for this option. |
| AWS EC2 | 8 | $27,724$2.77 per 1M tokens | Price not verified yet. | Price not verified yet. |
Every offer and its price
Every listed offer for this GPU. Where a provider sells several instance sizes, the tables above use the cheapest size that fits the GPU count.
| Provider | Instance | Location | GPUs per instance | On demand | Spot | 1-year reserved |
|---|---|---|---|---|---|---|
| AWS EC2 | p6-b200.48xlarge | US East (N. Virginia), Linux | 8 | $113.9328per instance-hour | n/a | n/a |
| Lambda | 8x NVIDIA HGX B200 | Lambda Cloud | 8 | $7.69per GPU-hour | Not offered | Not offered |
| Lambda | 4x NVIDIA B200 SXM6 | Lambda Cloud | 4 | $7.79per GPU-hour | Not offered | Not offered |
| Lambda | 2x NVIDIA B200 SXM6 | Lambda Cloud | 2 | $7.89per GPU-hour | Not offered | Not offered |
| Lambda | 1x NVIDIA B200 SXM6 | Lambda Cloud | 1 | $7.99per GPU-hour | Not offered | Not offered |
| Runpod Secure Cloud | B200 | Runpod Secure Cloud | Any | $7.99per GPU-hour | Not offered | Not offered |
| Runpod Community Cloud | B200 | Runpod Community Cloud | Any | $5.98per GPU-hour | Not offered | Not offered |
Cost per million tokens uses 11,436 tokens per second per GPU. MLPerf Inference v6.1, Offline scenario, Closed Division, gpt-oss 120B: 91,487 tokens/sec on CoreWeave B200 SXM (8x B200-SXM-180GB). Divided by 8 GPUs. A maximum-throughput benchmark result for one model, not a guarantee for other models or traffic patterns.
Your traffic looks different? Enter your own figures in the calculator.
Related pages
Other GPUs
Sources
Sources: aws.amazon.com (checked ); b0.p.awsstatic.com (checked ); developer.nvidia.com (checked ); lambda.com (checked ); runpod.io (checked )
Prices are list prices as each provider publishes them, used unchanged. How the figures are calculated: methodology.