Overview
orl is the OpenRelay command-line tool. Log in with an API key, deploy GPU VMs, open SSH sessions, and drive the whole API from your terminal or an agent.
orl is the OpenRelay command-line tool. It wraps the same REST API these docs
describe, so anything you can do over HTTPS you can do in one line from a
terminal: launch a GPU VM, SSH into it, check your balance, manage keys and
clusters, or serve the whole API to an agent.
What you get, and what it costs
orl deploy gives you a GPU VM you SSH into, billed per hour while it runs. You
start with free credit and no card required, so you can try it before adding a
payment method. Check live rates any time with orl catalog pricing get, and
stop billing with orl vms stop or orl vms terminate.
Two products, one CLI
Want an OpenAI-compatible API endpoint to call from your app? That is the
Inference API, a hosted gateway you reach over HTTPS. No CLI
command sends inference requests; you point an OpenAI or Anthropic client at it.
Want your own GPU box or container to run whatever you like? Keep reading: that
is what orl deploys.
Install
brew install openrelayinc/tap/orlThe Homebrew cask is signed and clears macOS quarantine for you, so orl runs
without a Gatekeeper prompt.
Download the .deb or .rpm for your architecture from the
latest release and install it:
# Debian / Ubuntu
sudo dpkg -i orl_*_linux_amd64.deb
# Fedora / RHEL
sudo rpm -i orl_*_linux_amd64.rpmGrab the archive for your platform from the
latest release, extract it, and
put orl on your PATH.
macOS: clear the quarantine flag
A binary you download directly (not through Homebrew) is quarantined by Gatekeeper. Clear it once after extracting:
xattr -dr com.apple.quarantine ./orlVerify the install and see the version:
orl version