Over 80,000,000 Cloud Servers Launched

Over 80,000,000 Cloud Servers Launched

Cloud Compute Cloud GPU Bare Metal File System Object Storage Block Storage Managed Databases CDN Serverless Kubernetes Container Registry Direct Connect Load Balancers

Regions Advanced Network Control Panel Operating Systems Upload ISO

Industry Cloud One-Click Deployment Use Cases

Browse Apps Become a Vendor

FAQ Developers / APIs Vultr Docs Server Status Bug Bounty Promotions Solution Partners Start-Up Programs

Our Team News Brand Assets Referral Program Creator Program Careers SLA Legal Vultr Trust Center Contact Your Privacy Choices Subprocessors Accessibility

Contact Sales Sign Up

© Vultr 2026 | VULTR is a registered trademark of The Constant Company, LLC.

Terms of Service AUP DMCA Privacy Policy Cookie Policy

Pricing

Dashboard Contact Sales

⌘K

Blogs
Discover
Docs
Community

Inference Cookbook

Benchmark Methodology Benchmark Results Overview

Troubleshooting Production Deployment

Benchmarks | Vultr Docs

Docs
Inference Cookbook
Cookbook for CUDA
Benchmarks

Benchmarks

Benchmark methodology and performance results for LLM inference workloads on NVIDIA HGX B200 GPUs.

Articles

Benchmark Methodology

11 March 2026

Benchmark Results Overview

11 March 2026

Tech Talk: Running a Self-Improving AI Agent with Hermes on Vultr

From The Blog

Accelerate OpenFold3 with AMD and Vultr

September 14, 2026

Why CPUs Are the Workhorse of Agentic AI Infrastructure

September 9, 2026

Run Your Own Inference Stack on Vultr with Modelplane

September 3, 2026