GPU catalog

Rent cloud GPUs for AI & ML

Choose from NVIDIA's most powerful GPUs. Deploy in minutes with built-in auto-failover, no egress fees, and per-minute billing.

Get started free Pricing calculator

From $0.25/hr
Per-minute billing
No egress fees
Automatic failover

NVIDIA RTX 4090

VRAM: 24GB GDDR6X
CUDA cores: 16,384
Architecture: Ada Lovelace
FP32: 82.6 TFLOPS

LLM inference, Stable Diffusion, real-time AI applications

NVIDIA RTX 3090

VRAM: 24GB GDDR6X
CUDA cores: 10,496
Architecture: Ampere
FP32: 35.6 TFLOPS

Cost-effective inference, fine-tuning small models, batch processing

NVIDIA RTX Pro 6000

VRAM: 96GB GDDR7
CUDA cores: 21,760
Architecture: Blackwell
FP32: 105.1 TFLOPS

Large model inference, 96GB VRAM at the lowest price, cost-effective training

NVIDIA H100 SXM

VRAM: 80GB HBM3
CUDA cores: 16,896
Architecture: Hopper
FP32: 67 TFLOPS

Large model training, multi-GPU workloads, enterprise AI

NVIDIA A100

VRAM: 40GB HBM2e
CUDA cores: 6,912
Architecture: Ampere
FP32: 19.5 TFLOPS

Mixed-precision training, HPC, large batch inference

NVIDIA H200

VRAM: 141GB HBM3e
CUDA cores: 18,432
Architecture: Hopper
FP32: 67 TFLOPS

Massive LLMs, 1.8x more memory than H100, next-gen training

NVIDIA RTX 5090

VRAM: 32GB GDDR7
CUDA cores: 21,760
Architecture: Blackwell
FP32: 105.1 TFLOPS

Next-gen inference, successor to RTX 4090 with 33% more VRAM

Not sure which GPU to pick?

Our team can help you choose the right GPU for your workload. Start free — no credit card required.

Start free Talk to us