Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.ornn.com/llms.txt

Use this file to discover all available pages before exploring further.

Ornn is a GPU compute exchange that connects you with high-performance GPU clusters for AI training and inference workloads. Browse available capacity on the marketplace, place bids at your target price, and access reserved compute through a virtual machine or direct Bare Metal host — all from a single platform.

Quick Start

Sign up, get approved, and reserve your first GPU cluster in minutes.

Sign In

Learn how to sign in with magic link, Google, or GitHub.

Browse the Marketplace

Find GPU capacity by type, facility, and availability.

Reserve Compute

Pick dates, set a bid price, and lock in your GPU allocation.

Access Your Compute

Connect to your reserved GPUs via VM or direct Bare Metal SSH.

Manage Your Portfolio

Track reservations, monitor GPU usage, and view invoices.

How Ornn works

1

Sign up and get approved

Create an account with your work email. The Ornn team reviews your company profile and approves your tenant account, typically within one business day.
2

Browse marketplace listings

Once approved, explore available GPU clusters filtered by GPU type (H100, A100, etc.) and facility location.
3

Place a bid and reserve

Select a deployment, choose your start and end dates, set a GPU count, and submit a bid price per GPU-hour. Accepted bids become confirmed reservations.
4

Choose your access mode

Pick VM for a ready-to-use virtual machine with the Ornn base image (or your own approved image), or Bare Metal for direct host access with maximum performance.
5

Run your workloads

SSH into your compute environment and start training. Monitor GPU-hour usage and job history from your portfolio dashboard.

Key concepts

TermMeaning
TenantYour organization’s account on Ornn
Inventory / ListingA GPU cluster available for reservation
BidAn offer to reserve capacity at a stated price per GPU-hour
ReservationA confirmed, time-bounded GPU allocation
PortfolioYour collection of active and past reservations
VM accessAccess your reservation through a virtual machine
Bare Metal accessDirect SSH to the physical host for maximum performance