tools

Lambda GPU Cloud

Lambda GPU Cloud provides on-demand GPU instances, connected GPU clusters, and dedicated superclusters for AI training and inference.

Lambda GPU Cloud lets researchers, developers, and companies launch NVIDIA GPU instances for training, fine-tuning, serving, and agent workloads. It offers self-serve instances, 1-Click Clusters with interconnected GPUs, and larger reserved infrastructure.

Instances support UI, API, and CLI workflows, with Lambda Stack providing tools such as PyTorch and CUDA. Usage is billed by GPU-hour, and larger clusters or reserved capacity require contacting Lambda; applicable taxes may apply.

Features

Launch 1-8 NVIDIA GPU instances on demand
Provision 16 to 2,000+ GPU 1-Click Clusters
Use the Lambda Cloud API, web UI, or CLI
Run Linux-based, GPU-backed virtual machines
Use Lambda Stack with PyTorch and CUDA pre-installed
Monitor GPU, memory, and network performance
Attach storage for datasets, checkpoints, and outputs
Use workspaces for resource grouping and access control

Use cases

Train and fine-tune machine-learning models
Serve inference workloads and AI agents
Prototype workloads on single or multi-GPU instances
Run distributed training on interconnected clusters
Benchmark models and GPU configurations
Separate development, staging, and production resources

Pros

Cons

Latest updates

2026-11-04 Introducing identity and access management for Lambda Cloud
Role-based access control with roles, groups, and service identities is now available in the Lambda Cloud console.
2026-10-13 Lambda to integrate open-source NVIDIA Run:ai with its Managed Kubernetes Service
Lambda is working to make open-source NVIDIA Run:ai a first-class option on Lambda Kubernetes Service (LKS).

Capabilities

Command line — “Use UI, API, or CLI” source
API — “Automate with the Lambda Cloud API to create, stop, and restart instances from your CLI, CI/CD, or orchestration scripts.” source
Runs models for you — “Train, fine-tune, and serve models on 1 to 8 NVIDIA GPU instances” source

Get it

Documentation

Security

SOC 2 Type II — “SOC 2 Type 2” source
SOC 2 — “SOC 2 Type 2” source
ISO 27001 — “ISO/IEC 27001:2022” source

Pricing

Prices checked: 2026-11-05

NVIDIA HGX B200 — 16 GPUs

$9.86/GPU/HR

2 weeks – 1 year

NVIDIA HGX B200 — 64 GPUs

$9.36/GPU/HR

2 weeks – 1 year

NVIDIA HGX B200 — 256+ GPUs

$8.87/GPU/HR

2 weeks – 1 year

NVIDIA H100 — 16 GPUs

$6.16/GPU/HR

2 weeks – 1 year

NVIDIA H100 — 64 GPUs

$5.85/GPU/HR

2 weeks – 1 year

NVIDIA H100 — 256 GPUs

$5.54/GPU/HR

2 weeks – 1 year

8x NVIDIA HGX B200

$6.69/GPU/HR

180 GB VRAM/GPU
208 vCPUs
2900 GiB RAM
22 TiB SSD storage

8x NVIDIA HGX H100

$3.99/GPU/HR

80 GB VRAM/GPU
208 vCPUs
1800 GiB RAM
22 TiB SSD storage

Official website