OpenRouter¶

OpenRouter speaks the OpenAI Chat Completions format at https://openrouter.ai/api/v1. That path is not among Privyx's default routes, so add it. Without the route, requests are forwarded unmasked.

Configure¶

# openrouter.yaml
proxy:
  routes:
    /api/v1/chat/completions: openai

privyx proxy -c openrouter.yaml --upstream https://openrouter.ai

Connect a client¶

The client keeps OpenRouter's /api/v1 in its base URL and its OpenRouter key, which Privyx relays:

import os

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:8000/api/v1",
    api_key=os.environ["OPENROUTER_API_KEY"],
)

reply = client.chat.completions.create(
    model="openai/gpt-4o-mini",
    messages=[{"role": "user", "content": "Write a short greeting to alice@example.com"}],
)
print(reply.choices[0].message.content)

Optional headers such as HTTP-Referer pass through unchanged.

Good to know¶

proxy.routes replaces the default map. If the same proxy should also serve the standard paths, list them too; see Routes.
Other OpenRouter paths, such as /api/v1/models, are forwarded as sent. Set proxy.passthrough_unknown: false to refuse anything that is not masked.
To check the route, run privyx audit tail after a request: session.transform appears only for masked paths, and proxy.request shows schema=openai.