Streaming

SSE streaming for lower time-to-first-token and a better UX.

Enabling it

Set "stream": true in the request body. The response is text/event-stream, delivering delta.content chunk by chunk and terminating with [DONE].

# pip3 install openai
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ.get("AIGATEWAY_API_KEY"),
    base_url="https://api.aigateway.ltd/v1",
)
resp = client.chat.completions.create(
    model="auto",           # or pin one: "deepseek-flash" / "qwen-max"
    messages=[{"role": "user", "content": "Hello!"}],
    stream=True,
)
for chunk in resp:
    print(chunk.choices[0].delta.content or "", end="")

Timeout guidance