LangChain Integration

LangChain can integrate with CoreRouter via OpenAI-compatible interfaces. Core configuration remains API Key, Base URL, and Model ID.

Python

Installation

pip install langchain-openai

Basic Usage

from langchain_openai import ChatOpenAI
import os

llm = ChatOpenAI(
    model="claude-sonnet-4-5",
    api_key=os.environ["COREROUTER_API_KEY"],
    base_url="https://api.corerouter.cloud/v1",
)

response = llm.invoke("Hello, please introduce yourself.")
print(response.content)

Streaming Output

for chunk in llm.stream("Introduce CoreRouter in three sentences."):
    print(chunk.content, end="", flush=True)

Chain Example

from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import ChatPromptTemplate
from langchain_openai import ChatOpenAI
import os

llm = ChatOpenAI(
    model="claude-sonnet-4-5",
    api_key=os.environ["COREROUTER_API_KEY"],
    base_url="https://api.corerouter.cloud/v1",
)

prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a professional translation assistant."),
    ("user", "Please translate the following to English: {text}"),
])

chain = prompt | llm | StrOutputParser()

print(chain.invoke({"text": "你好，世界"}))

JavaScript / TypeScript

Installation

npm install @langchain/openai @langchain/core langchain

Basic Usage

import { ChatOpenAI } from "@langchain/openai";

const llm = new ChatOpenAI({
  model: "claude-sonnet-4-5",
  apiKey: process.env.COREROUTER_API_KEY,
  configuration: {
    baseURL: "https://api.corerouter.cloud/v1",
  },
});

const response = await llm.invoke("Hello, please introduce yourself.");
console.log(response.content);

Chain Example

import { StringOutputParser } from "@langchain/core/output_parsers";
import { ChatPromptTemplate } from "@langchain/core/prompts";
import { ChatOpenAI } from "@langchain/openai";

const llm = new ChatOpenAI({
  model: "claude-sonnet-4-5",
  apiKey: process.env.COREROUTER_API_KEY,
  configuration: {
    baseURL: "https://api.corerouter.cloud/v1",
  },
});

const prompt = ChatPromptTemplate.fromMessages([
  ["system", "You are a professional translation assistant."],
  ["user", "Please translate the following to English: {text}"],
]);

const chain = prompt.pipe(llm).pipe(new StringOutputParser());

console.log(await chain.invoke({ text: "你好，世界" }));

Models and Capabilities

• Regular chat: Select any available chat model.
• Tool Calling / Agent: Select models marked as supporting tool calling or Coding Agent in the console.
• Vision: Select models that support image input.
• RAG: Chat models and Embeddings models may need separate configuration; refer to console model list.
• Responses API: If the application or Agent depends on Responses API, first validate /v1/responses separately with a Model ID that supports Responses; don't only validate Chat Completions.

Embeddings

from langchain_openai import OpenAIEmbeddings
import os

embeddings = OpenAIEmbeddings(
    model="embedding-model-id",
    api_key=os.environ["COREROUTER_API_KEY"],
    base_url="https://api.corerouter.cloud/v1",
)

vector = embeddings.embed_query("CoreRouter documentation")
print(len(vector))

Common Issues

Legacy Parameter Names Don't Work

Different LangChain versions may use different parameter names: model / modelName, apiKey / openAIApiKey. If the example throws a type error, upgrade dependencies first, or adjust according to current package version type hints.

Agent Cannot Call Tools

First test tool calling with the same model using regular OpenAI SDK examples. If SDK doesn't return tool_calls either, the current model or channel doesn't support that capability.

RAG Query Results Abnormal

Confirm the same Embeddings model is used for document indexing and querying, and check vector dimensions, text chunking, and recall count.