Framework Integrations
Integrate KairosRoute with your favorite AI agent framework. Copy-paste code examples for every popular framework.
How It Works
KairosRoute is 100% OpenAI-compatible. Every framework connects via the OpenAI SDK or HTTP client. Simply:
- 1. Install the framework's LLM library
- 2. Point the base URL to
https://api.kairosroute.com/v1 - 3. Use your KairosRoute API key
- 4. Set
model="auto"to let KairosRoute intelligently route requests
LangChain / LangGraph
Use KairosRoute with LangChain's ChatOpenAI for simple tasks or LangGraph for multi-agent workflows.
Installation
pip install langchain-openai langgraphBasic Usage with ChatOpenAI
from langchain_openai import ChatOpenAI
# Initialize with KairosRoute endpoint
llm = ChatOpenAI(
base_url="https://api.kairosroute.com/v1",
api_key="kr-your-key",
model="auto" # KairosRoute picks the best model per request
)
# Simple invocation
response = llm.invoke("Summarize the key points of this document...")
print(response.content)
# With streaming
for chunk in llm.stream("Write a blog post about AI"):
print(chunk.content, end="", flush=True)
# Chain usage
from langchain.prompts import ChatPromptTemplate
prompt = ChatPromptTemplate.from_messages([
("system", "You are a helpful assistant."),
("user", "{input}")
])
chain = prompt | llm
result = chain.invoke({"input": "What is quantum computing?"})
print(result.content)LangGraph Multi-Agent
from langgraph.prebuilt import create_react_agent
from langchain_openai import ChatOpenAI
from langchain_core.tools import tool
# Create LLM with KairosRoute
llm = ChatOpenAI(
base_url="https://api.kairosroute.com/v1",
api_key="kr-your-key",
model="auto"
)
# Define some tools
@tool
def search_web(query: str) -> str:
"""Search the web for information"""
return f"Results for: {query}"
@tool
def calculate(expression: str) -> str:
"""Evaluate mathematical expressions"""
return str(eval(expression))
# Create agent
tools = [search_web, calculate]
agent = create_react_agent(llm, tools)
# Run agent
from langgraph.types import Command
result = agent.invoke({
"messages": [("user", "Research AI trends and calculate 2024 growth rate")]
})
print(result["messages"][-1].content)Pin Specific Models When Needed
# Use auto-routing for most calls (best cost/quality balance)
regular_llm = ChatOpenAI(
base_url="https://api.kairosroute.com/v1",
api_key="kr-your-key",
model="auto"
)
# Pin expensive models when you know you need them
expensive_llm = ChatOpenAI(
base_url="https://api.kairosroute.com/v1",
api_key="kr-your-key",
model="claude-opus-4-6" # Bypass routing for complex tasks
)
# Use in your chain
chain = prompt | expensive_llmAbout model="auto":
Requests with model="auto" are routed to the most cost-effective model that can handle your task. KairosRoute analyzes request complexity, required capabilities, and quality metrics in real-time. This typically saves 40-60% on LLM costs compared to using a single fixed model.
Learn more: LangChain OpenAI Docs | LangGraph Docs
Pro Tips
Use model="auto" for most calls
Let KairosRoute's intelligent router optimize each request. This typically saves 40-60% on LLM costs compared to using a single fixed model, with no quality loss.
Pin expensive models when needed
For tasks that absolutely require top-tier models (e.g., complex reasoning, creative writing), specify the model directly: model="claude-opus-4-6". You bypass routing but maintain control.
Routing overhead is minimal
The intelligent routing algorithm adds less than 50ms of latency per request on average. Network and provider latency typically dominate, making routing overhead negligible.
All responses follow OpenAI format
Streaming, tool use, JSON mode, and function calling all work transparently, regardless of which provider handles the request. Your code doesn't need to know which model was used.
Check response headers for routing info
Response headers include X-KairosRoute-Model (which model handled it) and X-KairosRoute-Savings (your cost savings). Use these for monitoring and debugging.
Next Steps
Get Your API Key
Create a KairosRoute account to generate an API key for your integrations.
Dashboard →