Python SDK

Use the official OpenAI Python SDK and set base_url to CloudService. Model IDs come from the models and pricing page, and every request field is listed in Chat Completions.

Set CLOUDSERVICE_BASE_URL to https://api.yourdomain.example/v1 for API credit or https://api.yourdomain.example/token/v1 for an API-token key. This is the OpenAI-compatible route; native Claude apps use their origin-only setup instead.

bash

# API-credit key:
export CLOUDSERVICE_BASE_URL="https://api.yourdomain.example/v1"
# API-token key:
# export CLOUDSERVICE_BASE_URL="https://api.yourdomain.example/token/v1"

Install

bash

pip install openai

Basic chat

client.py

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["CLOUDSERVICE_API_KEY"],
    base_url=os.environ["CLOUDSERVICE_BASE_URL"],
)

response = client.chat.completions.create(
    model="YOUR_AUTHENTICATED_MODEL_ID",
    messages=[{"role": "user", "content": "Write a haiku about fast inference."}],
)

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

Streaming

stream.py

stream = client.chat.completions.create(
    model="YOUR_AUTHENTICATED_MODEL_ID",
    messages=[{"role": "user", "content": "Stream a short greeting."}],
    stream=True,
)

for chunk in stream:
    delta = chunk.choices[0].delta.content
    if delta:
        print(delta, end="", flush=True)

Server-sent-event framing and safe disconnect handling are covered in Streaming. The same flow in other languages: JavaScript, cURL.