Nebius

AI Cloud (Self-service AI clusters): NVIDIA B200 GPU

NVIDIA B200 GPU delivers 180 GB HBM3e memory, up to 8 TB/s bandwidth, and $5.50 per GPU-hour—perfect for GenAI teams.

AI Cloud (Self-service AI clusters): NVIDIA H100 GPU

NVIDIA H100 GPU delivers FP8 mixed-precision support, 16 vCPUs and 200 GB RAM per GPU, and InfiniBand multi-GPU clustering—perfect for ML engineers training LLMs.

AI Cloud (Self-service AI clusters): NVIDIA H200 GPU

NVIDIA H200 GPU delivers 141 GB HBM3e, 4.8 TB/s memory bandwidth, and $3.50 per GPU‑hour on‑demand pricing—perfect for ML engineers and AI researchers.

AI Cloud (Self-service AI clusters): NVIDIA L40S GPU

NVIDIA L40S GPU delivers 48 GB VRAM, $1.55 per GPU-hour starting rate, and PCIe L40S with Intel Ice Lake support—perfect for ML engineers.

AI Cloud: Compute - NVIDIA H100 NVLink with Intel Sapphire Rapids GPU

Compute - NVIDIA H100 NVLink with Intel Sapphire Rapids GPU delivers 80 GB HBM3, 3.35 TB/s NVLink, and $2.118 per GPU-hour—perfect for ML engineers and HPC teams.

AI Cloud: Compute - NVIDIA H200 NVLink with Intel Sapphire Rapids GPU

Compute - NVIDIA H200 NVLink with Intel Sapphire Rapids GPU delivers NVIDIA H200 NVLink GPUs, Intel Xeon Platinum 8468 CPUs, and $2.668/GPU-hour pay-as-you-go—perfect for ML engineers.

AI Cloud: NVIDIA Blackwell Platforms

Blackwell Platforms delivers GB200 NVL72 72-GPU rack-scale clusters, HGX B200 8-GPU servers, and pay-as-you-go GPU hours—perfect for ML engineers and AI teams.

AI Cloud: NVIDIA L40S GPU with AMD

NVIDIA L40S GPU with AMD delivers 48 GB GDDR6 GPU memory, up to 864 GB/s bandwidth, and AMD EPYC CPU pairing—perfect for ML engineers.

AI Cloud: NVIDIA L40S GPU with Intel

NVIDIA L40S GPU with Intel delivers 48 GB GDDR6 GPU memory, up to 864 GB/s memory bandwidth, and selectable GPU counts in Nebius VM presets—perfect for ML engineers.