Lambda
GPU cloud offering self-serve NVIDIA B200, H100, A100 and GH200 instances, reservable 1-Click Clusters of 16 to 2,000+ GPUs, and larger private superclusters for training and inference.
About Lambda
Teams training or serving models need NVIDIA data-center GPUs, from a single instance to a multi-node cluster, without buying and running the hardware.
Who it’s for
- ML engineers and researchers who need on-demand H100 or B200 instances
- AI companies reserving multi-node GPU clusters for large training runs
When to consider it
Consider Lambda when you want published per-GPU hourly prices for current NVIDIA GPUs, self-serve instances from one to eight GPUs billed by the minute, and a path to reserved multi-node clusters of B200 or H100 GPUs as training grows.
Tradeoffs & limitations
- On-demand instances are priced per GPU hour and billed in one-minute increments while running, for example $4.29 for 1x H100 SXM, $3.99 per GPU on 8x H100 SXM and $6.99 for 1x B200, plus sales tax/VAT/GST. Instances are self-serve with first-come access.
- 1-Click Clusters start at 16 GPUs with 2-week to 1-year terms ($6.16 per H100 GPU hour or $9.86 per B200 GPU hour at 16 GPUs), are arranged through Lambda's team and billed weekly; filesystems are billed per GiB per month even when not mounted.
SOTA overview · Documentation-based assessment · Sources & review method
Updates
No updates shared yet.
Discussion
Newest firstAsk a question or share how you use Lambda.
Keep it helpful. Community rules
Loading discussion…
Sources & review method
Documentation-based assessment · Nov 2, 2026 · Prepared with AI assistance; not a hands-on benchmark.
Checked by SOTA · AI-assisted documentation review. Selection advice is our assessment; verify current requirements for your deployment.
Import history & original evidence
Lambda official website
GPU cloud offering self-serve NVIDIA B200, H100, A100 and GH200 instances, reservable 1-Click Clusters of 16 to 2,000+ GPUs, and larger private superclusters for training and inference.
Based on official pages and announcements checked on 2026-11-02. No hands-on test, performance benchmark or popularity ranking is claimed.