
Facts from Lambda’s own pricing and docs pages, September 2026 — 56 read from a quoted sentence on the page we archived.
Lambda rents GPUs from $0.69 per GPU-hour on demand, bills per minute, and asks no minimum commitment. Egress is free. Lambda does not publish an uptime SLA.
| On-Demand instances | 1-Click Clusters | |
|---|---|---|
| From | $0.69/GPU-hr (Quadro RTX 6000) | $5.54–6.16/GPU-hr (H100 by size) |
| Minimum term | None | 2 weeks to 1 year; 1 year+ by quote |
| Billing | Per minute, weekly invoice | Per GPU-hour, billed weekly per the reservation |
| Tenancy | VMs, 1× to 8× GPUs | Single-tenant clusters, 16 to 2,000+ GPUs, InfiniBand |
| Provisioning | Minutes; first-come access | Reservation reviewed and approved |
| Payment | Credit card ($10 pre-authorisation) | Invoice, due within ten days |
As of September 2026, the lowest published on-demand rate is $0.69 per GPU-hour for the NVIDIA Quadro RTX 6000 24GB. Lambda publishes an on-demand rate for 11 GPU models. Billing is per minute.
Lambda does not publish an uptime commitment on its own site.
No. Lambda states none on demand; 2 weeks on clusters.
No. Lambda states that egress is free.
Lambda states that signup is self-serve. On approval and identity checks it states: card pre-authorisation; purchases limited to listed countries.
Lambda states: service credits only; expire after 12 months.
Lambda lists 14 regions: US (9), Japan (2), India, Germany, Israel.
The contracting entity is Lambda, Inc. (Delaware).
Reviewed September 2026, from Lambda’s own site. 56 facts, each quoted from the page it was read on.