Instances | Lambda

Designed for builders

Launch NVIDIA HGX B200, H100, A100, or GH200 instances in minutes with self-serve, first-come access.

Spin up an instance and get straight to training or inference. No lengthy setup, no driver installs, just NVIDIA GPUs on Lambda Stack.

Choose 8x, 4x, 2x, or 1x GPU instances to fit a wide range of AI workloads, from POC to production.

Automate with the Lambda Cloud API to create, stop, and restart instances from your CLI, CI/CD, or orchestration scripts.

Transparent pricing with no egress fees.

	VRAM/GPU	vCPUs	RAM	STORAGE	PRICE/GPU/HR*
NVIDIA B200 SXM6	180 GB	208	2900 GiB	22 TiB SSD	$4.99
NVIDIA H100 SXM	80 GB	208	1800 GiB	22 TiB SSD	$2.99
NVIDIA A100 SXM	80 GB	240	1800 GiB	19.5 TiB SSD	$1.79
NVIDIA A100 SXM	40 GB	124	1800 GiB	5.8 TiB SSD	$1.29
NVIDIA Tesla V100	16 GB	88	448 GiB	5.8 TiB SSD	$0.55

	VRAM/GPU	vCPUs	RAM	STORAGE	PRICE/GPU/HR*
NVIDIA H100 SXM	80 GB	104	900 GiB	11 TiB SSD	$3.09
NVIDIA A100 PCIe	40 GB	120	900 GiB	1 TiB SSD	$1.29
NVIDIA A6000	48 GB	56	400 GiB	1 TiB SSD	$0.80

	VRAM/GPU	vCPUs	RAM	STORAGE	PRICE/GPU/HR*
NVIDIA H100 SXM	80 GB	52	450 GiB	5.5 TiB SSD	$3.19
NVIDIA A100 PCIe	40 GB	60	450 GiB	1 TiB SSD	$1.29
NVIDIA A6000	48 GB	28	200 GiB	1 TiB SSD	$0.80

	VRAM/GPU	vCPUs	RAM	STORAGE	PRICE/GPU/HR*
NVIDIA GH200	96 GB	64	432 GiB	4 TiB SSD	$1.49
NVIDIA H100 SXM	80 GB	26	225 GiB	2.75 TiB SSD	$3.29
NVIDIA H100 PCIe	80 GB	26	225 GiB	1 TiB SSD	$2.49
NVIDIA A100 SXM	40 GB	30	220 GiB	512 GiB SSD	$1.29
NVIDIA A100 PCIe	40 GB	30	225 GiB	512 GiB SSD	$1.29
NVIDIA A10	24 GB	30	226 GiB	1.3 TiB SSD	$0.75
NVIDIA A6000	48 GB	14	100 GiB	512 GiB SSD	$0.80
NVIDIA Quadro RTX 6000	24 GB	14	46 GiB	512 GiB SSD	$0.50

Turnkey performance: Full GPU access, zero throttling, and an optimized ML stack with essential tools like PyTorch and CUDA pre-installed via Lambda Stack.
Real-time visibility: Monitor GPU, memory, and network performance directly from the dashboard or API to catch bottlenecks before they slow training or inference.
Easy storage: Keep datasets, checkpoints, and outputs attached between sessions and scale up or down without re-uploading data or incurring egress fees.