Vultr
Simple cloud infrastructure worldwide
Vultr offers global access to the latest AMD and NVIDIA GPUs for AI/ML, AR/VR, high-performance computing, VDI/CAD, and more, available on demand either as virtual machines or bare metal across 32 worldwide data center regions.
Available GPUs
Hourly on-demand pricing. Click column headers to sort.
Prices last updated: March 23, 2026
GPU Model↑ | Memory↑ | GPUs↑ | vCPUs↑ | RAM↑ | Price / hr↑ |
|---|---|---|---|---|---|
| A100 PCIE | 40GB | 1x | — | — | $1.29/hr |
| A100 PCIE | 40GB | 4x | — | — | $1.29/hr |
| A100 SXM | 80GB | 1x | 12 | 120 GB | $2.60/hr |
| A100 SXM | 80GB | 8x | — | — | $1.49/hr |
| A16 | 64GB | 1x | 6 | 64 GB | $0.51/hr |
| A16 | 64GB | 2x | 12 | 128 GB | $0.51/hr |
| A16 | 64GB | 4x | 24 | 256 GB | $0.51/hr |
| A16 | 64GB | 8x | 48 | 496 GB | $0.51/hr |
| A16 | 64GB | 16x | 96 | 960 GB | $0.57/hr |
| A40 | 48GB | 1x | 24 | 120 GB | $1.86/hr |
| A40 | 48GB | 4x | — | — | $1.86/hr |
| A40 | 48GB | 1x | — | — | $1.71/hr |
| B200 | 192GB | 1x | — | — | $2.99/hr |
| B200 | 192GB | 8x | — | — | $2.99/hr |
| GB300 | 576GB | 4x | — | — | $2.99/hr |
| GH200 | 96GB | 1x | — | — | $1.99/hr |
| GH200 | 96GB | 1x | — | — | $2.99/hr |
| H100 | 80GB | 1x | — | — | $2.99/hr |
| H100 | 80GB | 8x | — | — | $2.99/hr |
| H100 | 80GB | 1x | — | — | $2.99/hr |
| HGX B300 | 288GB | 8x | — | — | $2.99/hr |
| L40S | 48GB | 1x | — | — | $1.67/hr |
| L40S | 48GB | 8x | — | — | $1.67/hr |
| L40S | 48GB | 1x | — | — | $1.67/hr |
| MI300X | 192GB | 1x | — | — | $1.85/hr |
| MI300X | 192GB | 8x | — | — | $1.85/hr |
| MI325X | 256GB | 1x | — | — | $2.00/hr |
| MI325X | 256GB | 8x | — | — | $2.00/hr |
| MI355X | 288GB | 1x | — | — | $2.59/hr |
| MI355X | 288GB | 8x | — | — | $2.59/hr |
Pros & Cons
Advantages
- Wide range of AMD and NVIDIA GPU options
- Extensive global network (32 data center regions)
- Both VM and bare metal deployment options
- Kubernetes and containerization support
- NVIDIA Preferred Cloud Partner status
- Simplified driver setup and licensing
- API and Terraform integration
Limitations
- Medium GPU availability compared to some specialized providers
- Less established in the GPU market compared to major hyperscalers
- Limited documentation specific to GPU workloads
Key Features
AMD and NVIDIA GPU Options
Access to diverse GPU options including AMD Instinct and NVIDIA Tensor Core GPUs
Global Availability
Deploy GPU resources across 32 cloud data center regions worldwide
Kubernetes Support
Vultr Kubernetes Engine for GPU-accelerated containerized workloads
Serverless Inference
Deploy and scale GenAI models quickly with Vultr Serverless Inference
Virtual Machines and Bare Metal
Choose between GPU-accelerated VMs or dedicated bare metal servers
Global Content Delivery
Accelerate content delivery across six continents with Vultr CDN
Compute Services
NVIDIA GPU Instances
Virtual machines and bare metal servers with NVIDIA GPUs
AMD GPU Instances
Computing infrastructure powered by AMD Instinct accelerators
Vultr Kubernetes Engine
Managed Kubernetes service for GPU-accelerated containerized applications
- GPU-accelerated Kubernetes clusters
- Global deployment options
- Simplified container orchestration
- Support for resource-intensive workloads
Vultr Serverless Inference
Deploy and scale GenAI models efficiently
- Quick deployment of AI models
- Efficient scaling
- Support for proprietary data or trained models
- Global acceleration
Availability & Support
Regions
32 global cloud data center regions across North America, South America, Europe, Asia, Africa, and Australia
Support
Documentation, community forums, support tickets, and dedicated customer support
Getting Started
- 1
Create an account
Sign up for a free Vultr account
- 2
Select GPU instance
Choose from AMD or NVIDIA GPU options based on your workload
- 3
Choose deployment type
Select between virtual machine or bare metal deployment
- 4
Configure and launch
Set up networking, storage, and security options before launching
Compare Providers
Find the best prices for the same GPUs from other providers
