GPU Models
Run AI, ML, and HPC workloads on powerful cloud GPUs—without limits or wasted spend.
NVIDIA H200
Optimize AI Workloads
The NVIDIA H200 Tensor Core GPU brings next-generation performance to cloud infrastructure, enabling advanced AI training, data analytics, and HPC with exceptional efficiency. Built on the Hopper architecture, the H200 leverages FP8 precision and the Transformer Engine to accelerate NLP, generative AI, and large-scale simulation workloads.
With 141GB of HBM3e memory and up to 4.8 TB/s bandwidth, the H200 delivers massive throughput for big data, scientific research, and enterprise-scale applications. Designed for multi-instance workloads, it ensures predictable QoS, scalability, and reliability, making it ideal for mission-critical, performance-intensive environments.
NVIDIA H100
Efficiency. Speed. Power.
The NVIDIA H100 Tensor Core GPU delivers breakthrough acceleration for AI training, inference, and HPC workloads with unmatched performance and scalability. Powered by the Hopper architecture, it introduces the Transformer Engine for optimized FP8 precision, providing up to 4,000 TFLOPS for deep learning, generative AI, and advanced scientific computing.
Equipped with 80GB of HBM3 memory and 3.35 TB/s bandwidth, the H100 enables faster simulations, complex model training, large-scale analytics, and real-time decision-making. With support for NVLink and MIG partitioning, it ensures parallel workload execution, resource efficiency, and enterprise-grade reliability across demanding cloud environments.
NVIDIA A100
Train. Infer. Scale.
The NVIDIA A100 Tensor Core GPU powers AI, data analytics, and HPC applications with versatile performance and scalability. Built on the Ampere architecture, it delivers up to 312 TFLOPS using Tensor Cores, optimized for deep learning, natural language processing, and scientific workloads requiring massive parallelism.
Featuring 80GB of HBM2e memory and 2 TB/s bandwidth, the A100 enables efficient model training, large-scale simulations, and big data workflows. With MIG partitioning, multiple isolated workloads run simultaneously, ensuring high utilization, flexibility, and enterprise-grade reliability across diverse computing environments.
NVIDIA A40
Visual Compute Power
The NVIDIA A40 Tensor Core GPU brings versatile acceleration to enterprise workloads, enabling AI training, content modeling, and advanced simulations with balanced performance and efficiency. Based on the Ampere architecture, the A40 delivers strong FP32 and Tensor Core throughput, making it suitable for inference, rendering, and large-scale data analysis.
With 48GB of GDDR6 ECC memory and high memory bandwidth, the A40 supports faster simulations, virtualization, and parallel workflows across industries. Optimized for data center deployments, it ensures consistent throughput, energy efficiency, and enterprise-grade reliability for diverse cloud applications.
NVIDIA L4
Universal AI Accelerator
The NVIDIA L4 GPU provides energy-efficient acceleration for AI inference, video processing, and enterprise workloads with balanced performance and cost efficiency. Built on the Ada Lovelace architecture, it is optimized for media streaming, computer vision, and lightweight machine learning applications at scale.
With 24GB of GDDR6 memory and high-bandwidth support, the L4 enables efficient parallel data processing, real-time analytics, and scalable deployment in modern data centers. Its low power profile and flexibility make it a reliable choice for enterprises requiring performance, scalability, and energy savings.
NVIDIA Tesla V100
Advanced AI Acceleration
The NVIDIA V100 Tensor Core GPU delivers proven performance for AI training, HPC, and enterprise workloads with reliable scalability. Powered by the Volta architecture, it offers Tensor Core acceleration with up to 125 TFLOPS, making it ideal for machine learning, data science, and scientific simulations.
With 32GB of HBM2 memory and 900 GB/s bandwidth, the V100 supports faster training cycles, large dataset processing, and parallel analytics. Optimized for data centers, it provides consistent throughput, energy efficiency, and dependable performance for a wide range of professional applications.
NVIDIA L40s
AI. Graphics. Power.
The NVIDIA L40S GPU delivers high-performance acceleration for AI, generative workloads, and enterprise applications with optimized compute and memory capabilities. Based on the Ada Lovelace architecture, it offers advanced Tensor Core and FP32 performance for deep learning, rendering, and large-scale simulations.
Equipped with 48GB of GDDR6 memory and high-bandwidth throughput, the L40S supports graphics-intensive applications, AI inference, and real-time data processing. Designed for virtualization and multi-instance compute, it ensures flexible scaling, workload reliability, and energy-efficient performance across diverse enterprise cloud environments.
NVIDIA A30
Enterprise AI Compute
The NVIDIA A30 Tensor Core GPU provides balanced acceleration for AI, data analytics, and HPC workloads with efficiency and scalability. Built on the Ampere architecture, it delivers strong Tensor Core and FP32 performance, making it ideal for NLP, training, and real-time analytics at scale.
Featuring 24GB of HBM2 memory and high-bandwidth throughput, the A30 accelerates large data processing, complex modeling, and enterprise AI applications. With MIG partitioning, multiple workloads can run concurrently with predictable QoS, ensuring high utilization, efficiency, and enterprise-ready reliability across diverse computing environments.