MyIP

Diese Seite ist noch nicht auf Deutsch übersetzt. Es wird das englische Original angezeigt.

LangChain

Pointing `ChatOpenAI` at our endpoint

LangChain has no MyIP-specific integration, and does not need one. Give ChatOpenAI our base URL and key — we implement the OpenAI-compatible shape it already speaks.

ItemValue
Base URLhttps://openrouter.myip.co.kr/api/v1
Environment variableMYIP_API_KEY
Example modelsgoogle/gemma-4-26b-a4b, lgai/exaone-4.0-32b

Python

bash
pip install langchain langchain-openai
export MYIP_API_KEY="sk-mo-v1-..."
python
import os
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser

llm = ChatOpenAI(
    model="google/gemma-4-26b-a4b",
    base_url="https://openrouter.myip.co.kr/api/v1",
    api_key=os.environ["MYIP_API_KEY"],
    temperature=0.7,
    default_headers={
        "HTTP-Referer": "https://example.com",
        "X-Title": "My LangChain App",
    },
)

prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a concise assistant."),
    ("human", "{question}"),
])

chain = prompt | llm | StrOutputParser()
print(chain.invoke({"question": "What is the capital of South Korea?"}))

JavaScript / TypeScript

bash
npm install @langchain/openai @langchain/core
export MYIP_API_KEY="sk-mo-v1-..."
typescript
import { ChatOpenAI } from '@langchain/openai';
import { ChatPromptTemplate } from '@langchain/core/prompts';
import { StringOutputParser } from '@langchain/core/output_parsers';

const llm = new ChatOpenAI({
  model: 'google/gemma-4-26b-a4b',
  apiKey: process.env.MYIP_API_KEY,
  temperature: 0.7,
  configuration: {
    baseURL: 'https://openrouter.myip.co.kr/api/v1',
    defaultHeaders: {
      'HTTP-Referer': 'https://example.com',
      'X-Title': 'My LangChain App',
    },
  },
});

const prompt = ChatPromptTemplate.fromMessages([
  ['system', 'You are a concise assistant.'],
  ['human', '{question}'],
]);

const chain = prompt.pipe(llm).pipe(new StringOutputParser());
console.log(await chain.invoke({ question: 'What is the capital of South Korea?' }));

Streaming

python
for chunk in llm.stream("Write a short poem."):
    print(chunk.content, end="", flush=True)
typescript
const stream = await llm.stream('Write a short poem.');
for await (const chunk of stream) {
  process.stdout.write(String(chunk.content));
}

You do not need to enable stream_options.include_usage — we always request it. To have LangChain fold the final usage chunk into its own totals, though, turn on stream usage collection:

python
llm = ChatOpenAI(
    model="lgai/exaone-4.0-32b",
    base_url="https://openrouter.myip.co.kr/api/v1",
    api_key=os.environ["MYIP_API_KEY"],
    stream_usage=True,
)

Reading cost and the generation id

LangChain hides response headers, so the reliable route to cost is the generation id in the response metadata.

python
msg = llm.invoke("Hello")

print(msg.usage_metadata)                  # token counts
gen_id = msg.response_metadata.get("id")   # 'gen-...'
python
import requests

def cost_krw(generation_id: str) -> float:
    res = requests.get(
        "https://openrouter.myip.co.kr/api/v1/generation",
        params={"id": generation_id},
        headers={"Authorization": f"Bearer {os.environ['MYIP_API_KEY']}"},
        timeout=10,
    )
    return res.json()["data"]["total_cost"]   # KRW

print(cost_krw(gen_id), "KRW")

Settlement finishes asynchronously, so a lookup immediately after a stream ends can be early; wait a moment and retry. All amounts are in Korean won (KRW).

Tool calling

python
from langchain_core.tools import tool

@tool
def get_weather(city: str) -> str:
    """Look up the current weather for a city."""
    return f"{city}: clear, 21C"

agent_llm = llm.bind_tools([get_weather])
msg = agent_llm.invoke("What is the weather in Seoul?")
print(msg.tool_calls)

The model has to support tool calling. Check that supported_parameters in the GET /api/v1/models response contains tools. See Tool calling.

What you cannot use

  • OpenAIEmbeddings — there is no embeddings endpoint; you get 404 not_supported.
  • LangChain's image and audio integrations — those endpoints do not exist here.

For a RAG pipeline, generate embeddings elsewhere (a local embedding model, for instance) and send only the generation step through our gateway. The full list is in Unsupported endpoints.

Related

Zuletzt aktualisiert am 05.09.2026