Introduction
The OpenRelay REST API — deploy GPU VMs and inference clusters, manage organizations and billing, and automate your infrastructure.
The OpenRelay API is the programmatic interface to the OpenRelay GPU cloud. Everything you can do in the dashboard — launch GPU VMs, run autoscaling inference clusters, manage SSH keys, top up your balance, subscribe to events — you can do over HTTPS with a single API key.
It is a predictable, resource-oriented REST API: JSON request and response bodies, standard HTTP verbs and status codes, and bearer-token authentication.
Base URL
All requests go to:
https://api.openrelay.incEvery endpoint is versioned under /v1. For example:
curl https://api.openrelay.inc/v1/me \
-H "Authorization: Bearer vl_your_api_key"Organization-scoped
Most resources live inside an organization. Endpoints that operate on a
collection are nested under /v1/orgs/{orgId}/…, and your API key is bound to
a single organization — so the orgId in the path must match the key's org.
Conventions
- Format — Requests and responses are
application/json. SendContent-Type: application/jsonon any request with a body. - Authentication — Every request carries
Authorization: Bearer vl_…. See Authentication. - Timestamps — ISO 8601 / RFC 3339 strings in UTC (e.g.
2026-06-10T12:00:00Z). - Money — Monetary amounts are integer cents unless noted otherwise.
- IDs — Opaque strings; don't parse them.
- Errors — A non-2xx status with a JSON
{ "error": "…" }body. See Errors. - Pagination — List endpoints are cursor-paginated. See Pagination.
Get started
Quickstart
From API key to your first running GPU VM in a few minutes.
Authentication
Create vl_ API keys and authenticate every request.
API Reference
Every endpoint, with schemas and an interactive playground.
Webhooks
Receive signed, retried event callbacks.
Built for tools and LLMs
These docs are machine-readable out of the box:
- The full spec is served at
/openapi.json— drop it into Postman, an SDK generator, or your editor. - Every page is available as Markdown — use the Copy Markdown button, or
append
.md-style content routes for any page. /llms.txtand/llms-full.txtexpose the whole site for AI assistants.