Browse every supported GPU family in the synchronized public catalog, not just the cheapest listings. Compare customer-facing machine-hour rates and regions before opening the Marketplace.
Latest successful catalog synchronization: . This is a synchronized snapshot, not a reservation or a guarantee of capacity at launch. Prices and availability may change.
Showing 23 of 23 GPU families
Available in synchronized catalog
Evaluate for machine learning training, inference, data science, and GPU-accelerated development.
Workloads: Training, Inference, Development, GPU computing
VRAM per GPU: 40 / 80 GB
Available regions: US West, Canada Central, EU East, EU West, US Central, US East
From USD 0.48 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for large-model training, high-throughput inference, and GPU-accelerated AI workloads.
Workloads: Training, Inference
VRAM per GPU: 80 / 94 GB
Available regions: EU East, US Central, US East, Canada Central, Asia-Pacific, EU West, US West
From USD 2.18 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for memory-intensive model training, inference, and GPU-accelerated AI workloads.
Workloads: Training, Inference
VRAM per GPU: 141 GB
Available regions: EU East, US East, Asia-Pacific, EU West, US West
From USD 3.95 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for AI development, image generation, rendering, and other GPU-accelerated workloads.
Workloads: Rendering & visualization, Development, Media & generation
VRAM per GPU: 32 GB
Available regions: Asia-Pacific, Canada Central, EU East, EU West, US Central, US East, US West, India South
From USD 0.53 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for large-model training and inference.
Workloads: Training, Inference
VRAM per GPU: 192 GB
Available regions: US East
From USD 8.79 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for memory-intensive AI training and inference.
Workloads: Training, Inference
VRAM per GPU: 288 GB
Available regions: EU West
From USD 9.89 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for rendering, visualization, and GPU compute.
Workloads: Rendering & visualization, GPU computing
VRAM per GPU: 48 GB
Available regions: Canada Central, EU West
From USD 0.65 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for inference and media processing.
Workloads: Inference, Media & generation
VRAM per GPU: 24 GB
Available regions: EU East, EU West, US East
From USD 0.37 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for visual computing, rendering, and inference.
Workloads: Inference, Rendering & visualization
VRAM per GPU: 48 GB
Available regions: Canada Central, US Central, EU West
From USD 0.86 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for inference, rendering, and AI development.
Workloads: Inference, Rendering & visualization, Development
VRAM per GPU: 48 GB
Available regions: EU West, US Central, Asia-Pacific
From USD 0.88 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for GPU computing and established machine learning workloads.
Workloads: Training, GPU computing
VRAM per GPU: 16 GB
Available regions: US West, EU West
From USD 0.10 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for GPU development and rendering.
Workloads: Rendering & visualization, Development
VRAM per GPU: 10 GB
Available regions: Asia-Pacific, EU West, US East, US West
From USD 0.13 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for AI experimentation, image generation, and rendering.
Workloads: Rendering & visualization, Development, Media & generation
VRAM per GPU: 24 GB
Available regions: Asia-Pacific, Canada Central, EU East, EU West, US Central, US East, US West
From USD 0.18 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for image generation, rendering, and inference development.
Workloads: Inference, Rendering & visualization, Development, Media & generation
VRAM per GPU: 16 GB
Available regions: Canada Central, EU West, US East
From USD 0.26 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for AI development, inference, and rendering.
Workloads: Inference, Rendering & visualization, Development
VRAM per GPU: 24 GB
Available regions: Asia-Pacific, Canada Central, EU East, EU West, US Central, US East, US West
From USD 0.43 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for rendering and GPU-accelerated development.
Workloads: Rendering & visualization, Development
VRAM per GPU: 16 GB
Available regions: Asia-Pacific, Canada Central, US East
From USD 0.31 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for visual computing and smaller GPU workloads.
Workloads: Rendering & visualization
VRAM per GPU: 16 GB
Available regions: EU East
From USD 0.27 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for rendering, visualization, and GPU development.
Workloads: Rendering & visualization, Development
VRAM per GPU: 20 GB
Available regions: EU East
From USD 0.31 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for memory-intensive rendering and AI development.
Workloads: Rendering & visualization, Development
VRAM per GPU: 48 GB
Available regions: US Central
From USD 0.72 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for GPU development, inference experiments, and rendering.
Workloads: Inference, Rendering & visualization, Development
VRAM per GPU: 16 GB
Available regions: EU West
From USD 0.13 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for rendering and memory-intensive GPU development.
Workloads: Rendering & visualization, Development
VRAM per GPU: 48 GB
Available regions: Canada Central, EU East, US Central
From USD 0.44 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for professional graphics and GPU development.
Workloads: Development
VRAM per GPU: 32 GB
Available regions: EU East
From USD 0.80 / machine-hour (1 GPU)
Available in synchronized catalog
Evaluate for large rendering scenes and memory-intensive AI workloads.
Workloads: Rendering & visualization
VRAM per GPU: 96 GB
Available regions: EU East, EU West, US Central, US East
From USD 2.19 / machine-hour (1 GPU)
These public listings do not yet have a dedicated family guide.
Available in synchronized catalog
EU East / Secure
USD 0.28 / machine-hour
Available in synchronized catalog
Canada Central / Secure
USD 0.30 / machine-hour
Available in synchronized catalog
EU East / Community
USD 0.26 / machine-hour
VRAM is memory per GPU; system RAM is separate. Multiple GPUs do not automatically form a single memory pool. Compare GPU count, region, capacity class, CPU, RAM, storage, and software compatibility against your workload.
Each family page lists available configurations from the same public JSON feed. Rates are USD for the whole listed machine per hour, not per GPU, and no currency conversion is applied. The feed selects the lowest customer price for each public model, GPU count, VRAM, region, and capacity class; it is not a machine inventory count.
For training, budget memory for weights, activations, and optimizer state. For inference, include context and concurrent requests. For rendering, check scene memory, renderer support, and licensing. No model name alone guarantees a particular throughput.
Review the exact configuration and authorized rate in the Marketplace before launch. Read the GPU cloud guide for environment, access, and lifecycle details.
A family guide helps you compare hardware; the deployment decision belongs to an exact offering. Use this checklist when moving from the directory to your workspace.
Match the full configuration. Check the listed GPU count and memory, then CPU, system RAM, storage, and compatible environment. GPU count describes devices inside one machine. Record your workload requirements before choosing a row in the GPU Marketplace.
Review authorization. Sign in, check available wallet balance, and review the approved rate for the selected offering. An active reservation reduces available balance; it is different from a completed usage charge. The billing guide explains reservations and usage records.
Prepare access and cleanup. Register your public SSH key, wait for provisioning, and verify the displayed connection fingerprint. Save required outputs before termination and review the resulting status and usage. Follow the instance lifecycle guide through completion.
A family can remain in this directory when it has no synchronized available configuration. Its hardware guidance is not a capacity reservation. The catalog timestamp reports synchronization, while the Marketplace is where you review the exact offering before deployment.
For automated comparisons, use the public GPU availability API. Its public rows describe grouped configurations and customer rates; they do not expose your account balance, private instances, or SSH credentials.