Skip to content
OpenRelay is in early access, and the /v1 API is stable. New capabilities ship in the changelog.
OpenRelay Platform

Production inference starts here.

OpenAI-compatible inference across a distributed global GPU network. Switch endpoints, keep your SDK, and deploy in minutes.

Start Building Get API key API Reference
chat.py
from openai import OpenAI

client = OpenAI(
    base_url="https://inference.openrelay.inc/v1",
    api_key="vl_your_api_key",
)

# see /docs/inference/models for available model ids
completion = client.chat.completions.create(
    model="openrelay/gpt-oss-120b",
    messages=[
        {"role": "user", "content": "Explain MoE in one line."},
    ],
)

print(completion.choices[0].message.content)
SDKs

Keep the SDK you already have

OpenRelay speaks the OpenAI and Anthropic wire formats. Change the base URL and API key; the official libraries and any OpenAI-compatible tool keep working. Two lines, not a rewrite.

OpenAIPoint base_url at OpenRelay. Same client.
AnthropicCall /v1/messages with the Anthropic SDK.
Pythonopenai and anthropic libraries, unchanged.
JavaScriptopenai-node and the Vercel AI SDK.
cURLPlain HTTP. No SDK required.
LangChainChatOpenAI with a custom base URL.
Read the SDK guide
client.ts
import OpenAI from "openai";

const client = new OpenAI({
  // the only two lines that change
  baseURL: "https://inference.openrelay.inc/v1",
  apiKey: process.env.OPENRELAY_API_KEY,
});
Documentation

Browse by stage

1

Get started

QuickstartChoose a model
2

Build

3

Operate

API KeysAuthenticationWebhooksBillingPricingErrors
Providers

Bring your GPUs online

Contribute GPU hardware to the network. Customer workloads run on your machines and you earn a share of the usage. Onboarding a node is one command.

Become a providerHardware requirements
bash
curl -fsSL \
  "https://api.openrelay.inc/v1/provider/bootstrap?token=vtk_..." \
  | sudo bash