Integration

Integration is one change: point base_url at metaproxy. Keep your SDK, request schema, and streaming behavior.

Your first request in three minutes.

1
Create a key
One key works with every available model.
2
Replace base_url
https://api.metaproxy.agency/v1. The request schema and streaming responses remain OpenAI-compatible.
3
Switch with the model field
Model choice stays in your product logic.

curl https://api.metaproxy.agency/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-opus-5",
    "messages": [{"role": "user", "content": "Hello"}]
  }'
# Switch models by changing only this value: model="..."

Your existing clients work as-is.

Any OpenAI-compatible client that supports a custom base_url can connect.

OpenAI Python SDK

pip install openai, then change base_url.

from openai import OpenAI
client = OpenAI(base_url="https://api.metaproxy.agency/v1", api_key="YOUR_API_KEY")

OpenAI Node SDK

npm i openai, then change baseURL.

import OpenAI from 'openai'
const client = new OpenAI({ baseURL: 'https://api.metaproxy.agency/v1', apiKey: 'YOUR_API_KEY' })

LangChain

Set ChatOpenAI(base_url=...).

from langchain_openai import ChatOpenAI
llm = ChatOpenAI(base_url="https://api.metaproxy.agency/v1", api_key="YOUR_API_KEY")

Cherry Studio / NextChat / LobeChat

Set the custom OpenAI endpoint to the gateway URL.

Custom OpenAI endpoint: https://api.metaproxy.agency/v1

curl / any HTTP client

POST https://api.metaproxy.agency/v1/chat/completions.

curl https://api.metaproxy.agency/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY"

Vercel AI SDK

createOpenAI({ baseURL }).

import { createOpenAI } from '@ai-sdk/openai'
const openai = createOpenAI({ baseURL: 'https://api.metaproxy.agency/v1' })

Moving from the official API takes two edits.

Client	Before	Now	Changes
Python SDK	OpenAI() points to the official endpoint	`base_url="https://api.metaproxy.agency/v1"`	2 lines
Node SDK	new OpenAI() points to the official endpoint	`baseURL: 'https://api.metaproxy.agency/v1'`	2 lines
LangChain	ChatOpenAI() points to the official endpoint	`base_url="https://api.metaproxy.agency/v1"`	2 lines
cURL	https://api.openai.com/v1/chat/completions	`https://api.metaproxy.agency/v1/chat/completions`	2 edits

base_url must end in /v1.
Streaming and non-streaming requests are supported.
Create keys in the console. The full key is shown only once.

Three controls for every key.

Credit limit

Set a per-key cap to contain runaway usage.

Expiration

Give a key an expiration time and it disables automatically.

Model access

Restrict a key to selected models.

The gateway provides OpenAI-compatible chat/completions for streaming and non-streaming requests, and an Anthropic-compatible messages endpoint for clients that speak that protocol — including Claude Code, which can therefore address any model in the catalog, not only Anthropic ones. Other endpoints depend on what is currently available in the console.