An OpenAI-compatible API: what it is and why it is convenient

What “compatible” means

A compatible API accepts requests in the OpenAI Chat Completions format: a list of messages with system, user and assistant roles, plus model, temperature and stream parameters. It answers in the same format, streaming included.

That is why the official openai SDKs, LangChain, n8n and editors like Cursor and Continue work with it: they only need a different address and key.

Switching providers in two lines

from openai import OpenAI

# before: client = OpenAI()
client = OpenAI(base_url="https://tenzo-cloud.com/v1", api_key="tc-your-key")

What to watch out for

• Model names differ between providers, change the model parameter
• Not every provider supports the same extra parameters
• Error format: Tenzo Cloud uses the OpenAI one, an error object with message, type and code
• Check the limits: we limit concurrent requests, not tokens

Why the standard won

When one API got the most users, tool authors started writing integrations for it. New providers found it easier to support the familiar format than to ask everyone to rewrite code. Compatibility became the norm: switching models is now as easy as changing a database address.

A provider migration checklist

1. Move base_url and the key into environment variables
2. Check which models you need and change their ids
3. Run your main scenarios and tests on the new model
4. Set up retries for 429, a normal answer when streams are capped
5. Keep a quick way back to the previous provider