Lambda review
Updated September 28, 2026 · DeployCue Editorial Team · Facts checked September 26, 2026
We earn a commission when you sign up through links on this site. Partner status can affect where a brand appears in our rankings. Partners with a live offer in your country come first, then every other brand by score.
Our verdict
GPU hour from $0.69 sits above the site median of €0.63, so the headline rate is not the cheapest on the table.
Key facts
- Best for
- Teams that want current NVIDIA SXM parts with no egress charge.
- GPU models
- NVIDIA B200 SXM6, NVIDIA H100 SXM, NVIDIA H100 PCIe, NVIDIA A100 SXM, NVIDIA A100 PCIe, NVIDIA GH200, NVIDIA Tesla V100, NVIDIA A10, NVIDIA A6000, NVIDIA Quadro RTX 6000↗
- GPU hour from
- $0.69↗
- Billing granularity
- Per minute↗
- Egress fee
- Free - not charged for ingress or egress↗
- Interconnect
- NVIDIA Quantum-2 InfiniBand↗
- Compliance certifications
- SOC 2 Type II, ISO 27001, ISO 27017, ISO 27701, ISO 22301, CCPA↗
- Legal entity
- Lambda, Inc.↗
- Headquarters
- United States↗
- Founded
- 2012↗
- Support channels
- Email, Help-centre, Community↗
- Pricing model
- Hourly billing
- Official website
- lambda.ai
Facts checked September 26, 2026, 7 sources
Apps and profiles
Pricing
Lambda NVIDIA V100 (USD)
compute: $0.79 / hour
Lambda NVIDIA H100 (USD)
compute: $3.29 / hour
Lambda NVIDIA GH200 Grace Hopper (USD)
compute: $2.29 / hour
Lambda NVIDIA B200 (USD)
compute: $6.69 / hour
Lambda NVIDIA RTX A6000 (USD)
compute: $1.09 / hour
Lambda NVIDIA A100 80GB (USD)
compute: $2.79 / hour
Lambda NVIDIA A10 (USD)
compute: $1.29 / hour
Pros and cons
Pros
- Free - not charged for ingress or egress, so exporting checkpoints and datasets costs nothing extra
- Current accelerators in NVIDIA B200 SXM6, NVIDIA H100 SXM and NVIDIA H100 PCIe, covering both SXM and PCIe form factors
- Multi-GPU clusters use NVIDIA Quantum-2 InfiniBand for fast node-to-node training
- SOC 2 Type II, ISO 27001 and ISO 27017 cover the usual vendor security review
Cons
- GPU hour from $0.69 is above the site median of €0.63, so the headline rate is not the cheapest on the list
- Support runs on email, help centre and community only, with no phone number or live chat listed
- The billing model is hourly, so a reserved or committed rate is not part of the plan
Who it suits
- Teams fine-tuning and serving on current NVIDIA SXM parts
- Researchers who move large datasets between the cluster and object storage
- Buyers who need SOC 2 Type II and ISO 27001 in their vendor review
Who should look elsewhere
- Buyers who need AMD accelerators, since NVIDIA B200 SXM6, NVIDIA H100 SXM and NVIDIA H100 PCIe is all NVIDIA
- Teams wanting reserved or committed rates, as the model is hourly
- Workloads that need phone or live chat help, since email, help centre and community covers neither
Plans, egress and billing
GPU hour from $0.69 on hourly instances, metered at per minute. Egress is the standout line: Free - not charged for ingress or egress, so exporting checkpoints and datasets adds nothing to the bill. Billing is hourly on public cloud instances, with no reserved tier in the list. The site median entry rate sits at €0.63, so compare the specific card before committing.
Core features
The catalogue in NVIDIA B200 SXM6, NVIDIA H100 SXM and NVIDIA H100 PCIe spans current and older NVIDIA boards in both SXM and PCIe form, from the newest SXM parts back to older cards, plus lighter inference options. Multi-GPU nodes are linked with NVIDIA Quantum-2 InfiniBand. The lineup is NVIDIA only, with no AMD part in the list.
How Lambda compares
Lambda vs RunPod9.3 vs 9.5
Lambda vs ScalaHosting9.3 vs 8.3
Lambda vs Contabo9.3 vs 9.2
Lambda vs Kamatera9.3 vs 9.5
Lambda vs Cloudways9.3 vs 9.2
Lambda vs PlanetHoster9.3 vs 8.7
Lambda vs Acronis9.3 vs 9.9
Lambda vs Aruba9.3 vs 9.3
Lambda vs Decodo9.3 vs 9.1
Lambda vs Hostinger9.3 vs 9.1- Lambda vs MarsProxies9.3 vs 9.0
Lambda vs hide.me9.3 vs 8.7
Lambda vs Proton Drive9.3 vs 8.3
Frequently asked questions
What does a GPU instance cost on Lambda?+
GPU instances start at $0.69. Billing is hourly at per minute granularity, so short jobs stop paying quickly. The catalogue in NVIDIA B200 SXM6, NVIDIA H100 SXM and NVIDIA H100 PCIe spans top-end SXM boards down to older cards, so entry pricing and top-end pricing differ by a wide margin.
Is there a charge for moving data in and out?+
Lambda states Free - not charged for ingress or egress. That matters most for training pipelines that pull large datasets out of the platform, or for anyone serving models that ship results off the cluster. Storage and compute charges still apply as normal.
Which GPUs can I rent, and how are they connected?+
The listed NVIDIA B200 SXM6, NVIDIA H100 SXM and NVIDIA H100 PCIe cover current and older NVIDIA parts in both SXM and PCIe form. Multi-GPU training clusters use NVIDIA Quantum-2 InfiniBand for the fabric between accelerators, which matters for jobs that need nodes to talk at speed.
What compliance certifications does Lambda hold?+
Lambda lists SOC 2 Type II, ISO 27001 and ISO 27017 on its trust documentation. The contract is with Lambda, Inc., a US company headquartered in the United States, and the company was founded in 2012. Buyers with data residency clauses should check the terms against their own requirements.
How do I reach Lambda support?+
Support is offered through email, help centre and community. There is no phone number or live chat in that list, so teams running training jobs around the clock will lean on documentation and the community rather than a live agent.
Does Lambda use subscriptions or pay as you go?+
Lambda bills hourly and meters usage at per minute. It has been operating since 2012 under the name Lambda, Inc., so a new account starts with usage-based charges rather than a subscription.
Does Lambda offer AMD GPUs as well as NVIDIA?+
The listed NVIDIA B200 SXM6, NVIDIA H100 SXM and NVIDIA H100 PCIe are NVIDIA parts throughout, in SXM and PCIe form. There is no AMD accelerator in the list, so buyers who need Instinct hardware should look at a different provider.
Sources
- docs.lambda.ai/public-cloud/billing checked September 26, 2026
Backs: Billing granularity, Egress fee
“ODC prices instances by hourly usage and bills in one-minute increments.”
- lambda.ai checked September 26, 2026
Backs: Interconnect
“Run on single-tenant NVIDIA GB300 NVL72 clusters with NVIDIA Quantum-2 InfiniBand for ultimate security and performance.”
- lambda.ai/about checked September 26, 2026
Backs: Founded
“Lambda is an AI-only company founded by ML engineers in 2012.”
- lambda.ai/legal/terms-of-service checked September 26, 2026
Backs: Headquarters, Legal entity
“or by writing to us at Lambda, Inc. 2510 Zanker Rd., San Jose, CA 95131.”
- lambda.ai/pricing checked September 26, 2026
Backs: GPU hour from, GPU models, Pricing model
“NVIDIA Quadro RTX 6000 | 24 GB | 14 | 46 GiB | 512 GiB SSD | $0.69”
- lambda.ai/support checked September 26, 2026
Backs: Support channels
“Open a ticket, engage with our community, and get self-service help for Lambda products and services.”
- lambda.ai/trust checked September 26, 2026
Backs: Compliance certifications
“SOC 2 Type II attestation. ISO 27001, 27017, 27701, and 22301 certified.”
Compiled from 7 source pages, checked September 26, 2026. We did not open an account.