We may earn a commission if you buy through links on this page. It never changes a score or a ranking. How we make money
Pricing is quote-based.
Lambda Cloud has no free plan or free trial listed. Self-serve GPU instances are billed per GPU-hour, with published rates from $0.69 for NVIDIA Quadro RTX 6000 to $7.99 for a NVIDIA B200 SXM6 (2-GPU listing). Production cluster rates are also per GPU-hour and vary by GPU count. The pricing page lists no monthly or annual billing. Rates checked November 6, 2026, are scheduled to take effect November 16; current rates are available after login until then.
01Every plan, line by line
| Plan | Billed annually | Billed monthly | Built for |
|---|---|---|---|
| NVIDIA HGX B200 (16 GPUs) | — | Quote | Production GPU clusters |
| NVIDIA HGX B200 (64 GPUs) | — | Quote | Production GPU clusters |
| NVIDIA HGX B200 (256+ GPUs) | — | Quote | Large production GPU clusters |
| NVIDIA HGX B200 (16+ GPUs, 1 year+) | — | Quote | Reserved GPU capacity |
| NVIDIA H100 (16 GPUs) | — | Quote | Production GPU clusters |
| NVIDIA H100 (64 GPUs) | — | Quote | Production GPU clusters |
| NVIDIA H100 (256 GPUs) | — | Quote | Large production GPU clusters |
| NVIDIA H100 (16+ GPUs, 1 year+) | — | Quote | Reserved GPU capacity |
| NVIDIA HGX B200 | — | Quote | Self-serve GPU workloads |
| NVIDIA HGX H100 | — | Quote | Self-serve GPU workloads |
| NVIDIA HGX A100 (80 GB) | — | Quote | Self-serve GPU workloads |
| NVIDIA HGX A100 (40 GB) | — | Quote | Self-serve GPU workloads |
| NVIDIA Tesla V100 | — | Quote | Self-serve GPU workloads |
| NVIDIA B200 SXM6 (8-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA H100 SXM (8-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA A100 PCIe (8-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA A6000 (8-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA B200 SXM6 (4-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA H100 SXM (4-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA A100 PCIe (4-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA A6000 (4-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA B200 SXM6 (2-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA GH200 | — | Quote | Self-serve GPU workloads |
| NVIDIA H100 SXM (2-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA H100 PCIe | — | Quote | Self-serve GPU workloads |
| NVIDIA A100 SXM | — | Quote | Self-serve GPU workloads |
| NVIDIA A100 PCIe (2-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA A10 | — | Quote | Self-serve GPU workloads |
| NVIDIA A6000 (2-GPU listing) | — | Quote | Self-serve GPU workloads |
| NVIDIA Quadro RTX 6000 | — | Quote | Self-serve GPU workloads |
Add-ons and overages
- Applicable sales tax, VAT, or GST may apply.
- Instance prices shown take effect November 16, 2026; current prices are available after login until then.
02What the plan grid doesn't tell you
GPU count and runtime
Rates are per GPU-hour. Estimate spend by multiplying the GPU rate by the GPUs used and the hours they run.
Taxes
Applicable sales tax, VAT, or GST may apply.
Reserved cluster quote
Rates for 1-year-plus NVIDIA HGX B200 and H100 reservations are not published; ask Lambda for a quote.
Effective-date change
Listed instance rates take effect November 16, 2026; current rates are available after login until then.
03Which plan should you buy?
| If you are… | Get | Why |
|---|---|---|
| Prototype or test workload | NVIDIA Quadro RTX 6000 | At $0.69 per GPU-hour, it is the lowest listed instance rate; match the GPU to workload requirements. |
| B200 production cluster | NVIDIA HGX B200 (16 GPUs) | The listed cluster rate is $9.86 per GPU-hour for the 16-GPU tier. |
| Long-term reserved capacity | NVIDIA H100 (16+ GPUs, 1 year+) | This is the listed H100 reservation option, but Lambda does not publish its price. |
04Is Lambda Cloud worth the money?
Lambda's published rates make GPU-hour costs visible across self-serve instances and production clusters, but the lack of a free tier means experimentation is paid compute. It may be worth paying for teams that need NVIDIA GPUs, dedicated clusters, or managed operations. Build a cost estimate from GPU count and runtime, then confirm the current rate, taxes, capacity, and any reservation quote with Lambda. For long-term H100 or B200 capacity, do not assume a discount: the 1-year-plus prices are not published.
Lambda Cloud pricing FAQ
What is the cheapest Lambda Cloud GPU?
The lowest listed self-serve rate is NVIDIA Quadro RTX 6000 at $0.69 per GPU-hour. Your total depends on how many GPUs you run and for how long. Applicable sales tax, VAT, or GST may apply.
Does Lambda Cloud offer a free tier?
The pricing information lists no free plan or free trial. Self-serve instances are charged per GPU-hour, so even a short test uses paid compute. Check the price shown after login for the current rate.
How are Lambda Cloud clusters priced?
The listed production cluster rates are per GPU-hour and depend on GPU model and cluster size. For example, NVIDIA HGX B200 (16 GPUs) is listed at $9.86 per GPU-hour, while NVIDIA H100 (16 GPUs) is $6.16.
Are reserved GPU prices published?
Lambda lists NVIDIA HGX B200 and H100 options for 16 or more GPUs with terms of one year or longer, but does not publish their prices. Ask Lambda for a quote and confirm the applicable rate and commitment terms.