Use the OpenAI Python SDK with InferGate
Configure the OpenAI Python SDK for InferGate, send a Responses request with gpt-5.5 and inspect text output without exposing your API key.
Install and configure
Use a supported Python version and a current OpenAI SDK release with Responses support. Install in an isolated virtual environment.
python -m venv .venv
. .venv/bin/activate
python -m pip install --upgrade openai
export INFERGATE_API_KEY="YOUR_INFERGATE_API_KEY"Send a complete response
import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["INFERGATE_API_KEY"],
base_url="https://api.useinfergate.com/v1",
)
response = client.responses.create(
model="gpt-5.5",
input="Reply with one sentence about API design.",
max_output_tokens=256,
)
print(response.output_text)Validate the result
Check response.status, response.model and response.usage as well as output text. Handle SDK exceptions in the caller. Do not log the client configuration or key. Follow the streaming guide for incremental text.
Python setup: common failure points
| Layer or outcome | Expected value or observation | Action |
|---|---|---|
| Key | os.environ["INFERGATE_API_KEY"] | Avoid hardcoding secrets |
| Client option | base_url | Include /v1 once |
| SDK | A release with client.responses support | Older SDK releases may lack Responses |
| Result | response.output_text and response.status | Do not treat empty output as task success |
Frequently asked questions
Why does Python report that responses is missing?
Check the installed package and interpreter in the active virtual environment. Install a current OpenAI SDK release with Responses support, then retry the minimal example.
How do I capture failures without leaking the key?
Catch the SDK exception at your application boundary and record sanitized error details, status and request ID. Do not dump the client object, environment or authorization header.
Can I stream with the Python SDK?
Yes, use the streaming guide for the verified Responses event pattern. The handler must check completion and close the stream when finished.
Continue your integration
Review prepaid pricing · Read API documentation · Get Started