GPU Instances

We are excited to announce expanded availability for L4 Instances, our most versatile and cost-effective GPU offering.

L4 GPUs are now available in a second Availability Zone in Paris (par-1, in addition to par-2), making it easier to build highly available inference infrastructure for your projects.

As a reminder, L4 GPU Instances are also available in the Warsaw region (waw-2).

Key features include:

• Nvidia L4 24 GB (Ada Lovelace architecture)
• 4th generation Tensor cores
• 4th generation RT cores (graphics capability)
• Available in multi-GPU 1, 2, 4 or 8 GPUs

Following the launch of our H100-SXM GPU Instances — delivering industry-leading conversational AI performance and accelerating large language models (LLMs) — we’re pleased to announce the availability of new 2-GPU and 4-GPU configurations.

With NVLink GPU-to-GPU communication, the 4-GPU option unlocks even greater possibilities and higher performance for your deployments. Now available in the Paris (PAR2) region.

Key features include:

• Nvidia H100 SXM 80GB GB (Hopper architecture)
• 4th generation Tensor cores
• 4th generation NVlink, which offers 900 GB/s of GPU-to-GPU interconnect
• Transformer Engine
• Available now in 2, 4 and 8 GPUs per VM (Additional stock deployments on-going)

We are excited to introduce our latest H100-SXM GPU Instances, delivering exceptional performance, scalability, and security for every workload. H100 uses breakthrough innovations based on the NVIDIA Hopper architecture to deliver industry-leading conversational AI, speeding up large language models (LLMs). Available in the Paris (PAR2) region.

Key features include:

• Nvidia H100 SXM 80GB GB (Hopper architecture)
• 4th generation Tensor cores
• 4th generation NVlink, which offers 900 GB/s of GPU-to-GPU interconnect
• Transformer Engine
• Available immediately in 8 GPUs (2 and 4 GPUs per VM coming soon)