LangChain - ModRouter

Endpoint

https://api.modrouter.online/langchain

ModRouter’s LangChain router combines OpenAI, Anthropic, GenAI, Bedrock, and Cohere route families under the /langchain prefix.

OpenAI Provider

import os
from langchain_openai import ChatOpenAI

base_url = os.environ["MODROUTER_BASE_URL"].rstrip("/")
api_key = os.environ["MODROUTER_API_KEY"]

llm = ChatOpenAI(
    model="openai/gpt-4o-mini",
    openai_api_base=f"{base_url}/langchain",
    openai_api_key=api_key,
)

print(llm.invoke("Hello").content)

Anthropic Provider

import os
from langchain_anthropic import ChatAnthropic

base_url = os.environ["MODROUTER_BASE_URL"].rstrip("/")
api_key = os.environ["MODROUTER_API_KEY"]

llm = ChatAnthropic(
    model="anthropic/claude-sonnet-4-5",
    anthropic_api_url=f"{base_url}/langchain",
    anthropic_api_key=api_key,
)

print(llm.invoke("Hello").content)

Bedrock Provider

For LangChain classes that use boto3 / Bedrock endpoint URLs, use:

https://api.modrouter.online/langchain

and dummy AWS credentials if the SDK requires signing credentials.

Notes

Choose the LangChain provider class that matches the request shape you want to send.
ModRouter’s /langchain endpoint is a compatibility prefix, not a LangChain runtime. Your chains, tools, and memory still run in your application.
If a LangChain integration sends an endpoint ModRouter does not implement, use a provider-specific ModRouter integration or passthrough.