OpenAI SDK - ModRouter

Endpoint

Use /openai as the SDK base URL for SDKs that append resource paths such as /chat/completions:

https://api.modrouter.online/openai

If a tool expects the base URL to already include /v1, use:

https://api.modrouter.online/openai/v1

Python

import os
from openai import OpenAI

base_url = os.environ["MODROUTER_BASE_URL"].rstrip("/")
api_key = os.environ["MODROUTER_API_KEY"]

client = OpenAI(
    base_url=f"{base_url}/openai",
    api_key=api_key,
)

response = client.chat.completions.create(
    model="openai/gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello"}],
)

print(response.choices[0].message.content)

JavaScript

import OpenAI from "openai";

const baseURL = process.env.MODROUTER_BASE_URL.replace(/\/$/, "");
const apiKey = process.env.MODROUTER_API_KEY;

const client = new OpenAI({
  baseURL: `${baseURL}/openai`,
  apiKey,
});

const response = await client.chat.completions.create({
  model: "anthropic/claude-sonnet-4-5",
  messages: [{ role: "user", content: "Hello" }],
});

console.log(response.choices[0].message.content);

Supported Route Groups

ModRouter’s OpenAI integration includes:

chat completions
text completions
responses
embeddings
speech and transcription
image generation, edits, and variations
videos where configured
files
batches
containers and container files
model listing
selected Azure OpenAI deployment-path compatibility
selected realtime HTTP, WebSocket, and WebRTC routes

The integration converts supported OpenAI request shapes into ModRouter requests. It is not a blanket proxy for every OpenAI endpoint. Use Passthrough for provider-native paths not covered by the integration.

Custom Headers

Pass ModRouter-specific headers through the SDK when needed:

response = client.chat.completions.create(
    model="openai/gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello"}],
    extra_headers={"x-bf-vk": os.environ["MODROUTER_API_KEY"]},
)

Async Inference

Integration routes support ModRouter’s async headers for inference requests when the async job store is configured:

job = client.chat.completions.create(
    model="openai/gpt-4o-mini",
    messages=[{"role": "user", "content": "Long task"}],
    extra_headers={"x-bf-async": "true"},
)

result = client.chat.completions.create(
    model="openai/gpt-4o-mini",
    messages=[{"role": "user", "content": "Long task"}],
    extra_headers={"x-bf-async-id": job.id},
)

Async is for inference routes, not file, batch, or container operations.

​Endpoint

​Python

​JavaScript

​Supported Route Groups

​Custom Headers

​Async Inference