What is an OpenAI-compatible API? Change one base_url to switch providers

An OpenAI-compatible API is one whose paths and request format match the OpenAI API, such as POST /v1/chat/completions. Tutujin API's OpenAI-compatible address is https://open.tutujin.com/v1. Code that already uses an OpenAI SDK needs two changes, base_url and api_key, to call GPT, Claude, Gemini, DeepSeek and every other model the catalog lists for that endpoint; set model to the model you want.

Why everyone supports the OpenAI format

The OpenAI format has the most users and the most tooling. The official Python and Node.js SDKs, LangChain, LlamaIndex, Dify, FastGPT and n8n all accept a custom base URL, so any service that speaks this format works with all of them out of the box. For a business, switching providers or models then needs no change to application code.

Endpoints on this site

"Verified" endpoints have been called for real on this site with billing checked. Endpoints "relayed as-is" pass requests to the model service in their original format, and parameters follow that model's official documentation.

Migrating in three steps

1. Create an API key in the console and pick a group for it.
2. Set base_url to https://open.tutujin.com/v1 and api_key to the new key.
3. Set model to a name from Models and pricing.

The official SDKs also read environment variables, so some projects need no code change at all:

export OPENAI_BASE_URL="https://open.tutujin.com/v1"
export OPENAI_API_KEY="sk-your-key"

Or set them explicitly in code:

import os
from openai import OpenAI

client = OpenAI(base_url="https://open.tutujin.com/v1", api_key=os.environ["TUTUJIN_API_KEY"])

resp = client.chat.completions.create(
    model="deepseek-v4.1-flash",
    messages=[{"role": "user", "content": "Explain context caching in three sentences"}],
)
print(resp.choices[0].message.content)

import OpenAI from 'openai'

const client = new OpenAI({ baseURL: 'https://open.tutujin.com/v1', apiKey: process.env.TUTUJIN_API_KEY })

const resp = await client.chat.completions.create({
  model: 'deepseek-v4.1-flash',
  messages: [{ role: 'user', content: 'Explain context caching in three sentences' }],
})
console.log(resp.choices[0].message.content)

Change only model to switch vendors, for example to claude-sonnet-5-5 or gemini-3.8-flash.

Compatible is not identical

• Parameter support varies. The range of temperature, response_format, tools and reasoning_effort are supported to different degrees by non-OpenAI models. An unsupported parameter may be ignored or may return a 400.
• Some response fields differ. The basic structure is the same, but vendor-specific details such as reasoning content or cached tokens depend on the model, both in field name and in whether they are returned.
• Use this site's model names. The same model can be named differently on different platforms; the model hub is the reference.

So after switching models, check parameters and responses with a few requests before sending real traffic.

Compatible or native?

All three use the same key and the same prices.

Tool setup guides

• OpenAI SDK
• Anthropic SDK
• LangChain and LlamaIndex
• Dify, FastGPT and n8n

Header formats and authentication options are in Authentication and base URL.