Connect with Python.

Install and configure

Install the current official SDK. Use a Python version supported by that SDK release.

python -m pip install --upgrade openai

Set your key and exact model ID in the terminal. Replace both placeholders. The environment variable names below belong to this guide; the client receives them explicitly.

export PRISMROUTE_API_KEY="YOUR_PRISMROUTE_API_KEY"
export PRISMROUTE_MODEL="YOUR_MODEL_ID"

$env:PRISMROUTE_API_KEY="YOUR_PRISMROUTE_API_KEY"
$env:PRISMROUTE_MODEL="YOUR_MODEL_ID"

Send a Chat Completions request

Save this as example.py. The base_url option is the API root, not the full /chat/completions URL. Use a model listed as supporting Chat Completions.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["PRISMROUTE_API_KEY"],
    base_url="https://codex.aicpr.icu/v1",
    timeout=60.0,
    max_retries=0,
)

completion = client.chat.completions.create(
    model=os.environ["PRISMROUTE_MODEL"],
    messages=[{"role": "user", "content": "Hello, PrismRoute!"}],
)
print(completion.choices[0].message.content or "")

python example.py

The example uses a 60-second timeout and disables automatic retries for the first connection check. For a production application, choose an explicit timeout and a bounded retry policy suitable for its work.

Read streamed text

For a model that supports streaming, read each text delta as it arrives. Some chunks carry metadata instead of text; the example handles empty choices and missing content.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["PRISMROUTE_API_KEY"],
    base_url="https://codex.aicpr.icu/v1",
    timeout=60.0,
    max_retries=0,
)

stream = client.chat.completions.create(
    model=os.environ["PRISMROUTE_MODEL"],
    messages=[{"role": "user", "content": "Hello, PrismRoute!"}],
    stream=True,
)
for chunk in stream:
    if chunk.choices:
        print(chunk.choices[0].delta.content or "", end="", flush=True)
print()

A streaming connection can fail after some text has arrived. Preserve that partial output and inspect the error before deciding whether to retry the request.

Use Responses when the model supports it

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["PRISMROUTE_API_KEY"],
    base_url="https://codex.aicpr.icu/v1",
    timeout=60.0,
    max_retries=0,
)

response = client.responses.create(
    model=os.environ["PRISMROUTE_MODEL"],
    input="Hello, PrismRoute!",
)
print(response.output_text)

Responses uses input and returns typed output items. The SDK’s output_text helper joins text output. Its streaming events differ from Chat Completions deltas; do not reuse the streaming loop above for a Responses stream.

Check your integration

Confirm that the generated request goes to codex.aicpr.icu, includes /v1 only once and uses the exact model ID. If the request fails, keep its status code and a redacted error message for troubleshooting.