# 🦜️🔗 LangChain Azure
This repository contains the following packages with Azure integrations with LangChain:
- [langchain-azure-ai](https://pypi.org/project/langchain-azure-ai/)
- [langchain-azure-compute](./libs/azure-compute)
- [langchain-azure-cosmosdb](https://pypi.org/project/langchain-azure-cosmosdb/)
- [langchain-azure-dynamic-sessions](https://pypi.org/project/langchain-azure-dynamic-sessions/)
- [langchain-sqlserver](https://pypi.org/project/langchain-sqlserver/)
- [langchain-azure-postgresql](https://pypi.org/project/langchain-azure-postgresql/)
- [langchain-azure-storage](https://pypi.org/project/langchain-azure-storage/)
**Note**: This repository will replace all Azure integrations currently present in the `langchain-community` package. Users are encouraged to migrate to this repository as soon as possible.
## LangGraph on Azure
Build with Azure models using the [LangChain and LangGraph development guide](https://aka.ms/azureai/langchain), or bring an existing compiled graph to Microsoft Foundry with the [LangGraph hosting guide](https://learn.microsoft.com/azure/foundry/how-to/develop/langchain-hosted-agents).
The [hosted-agent samples](./samples/hosting/langgraph-hosted-agents/README.md#start-here) cover local runs and deployment, including [hosting an existing graph without changing its implementation](./samples/hosting/langgraph-hosted-agents/responses/09_run/README.md).
For durable graph state, use [FoundryCheckpointSaver backed by Foundry State Store](./libs/azure-ai/README.md#durable-graph-state) in Foundry-hosted agents, or the [Azure Cosmos DB checkpointer](./libs/azure-cosmosdb/README.md#langgraph-checkpointer). Cosmos DB also provides a [long-term memory store](./libs/azure-cosmosdb/README.md#langgraph-store-long-term-memory).
## Samples
- [SDK samples](https://github.com/langchain-ai/langchain-azure/tree/main/samples) demonstrate SDK usage, including hosting, memory, and storage integrations.
- [SDK notebooks](https://github.com/langchain-ai/langchain-azure/tree/main/libs/azure-ai/docs) cover Foundry models, Agent Service, Content Safety, and Content Understanding.
- [Foundry deployment samples](https://github.com/microsoft-foundry/foundry-samples/tree/main/samples/python/hosted-agents/langgraph) demonstrate platform configuration and deployment with LangGraph.
# Quick Start with langchain-azure-ai
The `langchain-azure-ai` package uses the Microsoft Foundry family of SDKs and client libraries for Azure to provide first-class support of Microsoft Foundry capabilities in LangChain and LangGraph.
This package includes:
* [Microsoft Agent Service](./libs/azure-ai/langchain_azure_ai/agents)
* [Microsoft Foundry Models inference](./libs/azure-ai/langchain_azure_ai/chat_models)
* [Microsoft Foundry Content Safety](./libs/azure-ai/langchain_azure_ai/agents/middleware)
* [Microsoft Foundry Tools](./libs/azure-ai/langchain_azure_ai/tools)
* [Azure AI Search](./libs/azure-ai/langchain_azure_ai/vectorstores)
* [Azure AI Services tools](./libs/azure-ai/langchain_azure_ai/tools)
Here's a quick start example to show you how to get started with a chat model. For more details and tutorials see [Develop with LangChain and LangGraph and models from Azure AI Foundry](https://aka.ms/azureai/langchain).
### Install langchain-azure
```bash
pip install -U langchain-azure-ai
```
### Microsoft Foundry Models
Use any Foundry Model with OpenAI-compatible APIs:
We recommend the [Responses API](https://developers.openai.com/api/docs/guides/migrate-to-responses) when your model and endpoint support it. Set `use_responses_api=True` as shown below. If you prefer Chat Completions and your model and endpoint support it, set `use_responses_api=False` instead.
```python
from azure.identity import DefaultAzureCredential
from langchain_azure_ai.chat_models import AzureAIOpenAIApiChatModel
from langchain_core.messages import HumanMessage, SystemMessage
model = AzureAIOpenAIApiChatModel(
project_endpoint="https://{your-resource-name}.services.ai.azure.com/api/projects/{your-project}",
credential=DefaultAzureCredential(), # requires Azure AI Developer role. If using keys, use parameter `endpoint` instead of `project_endpoint`.
model="gpt-5", # your model deployment name
use_responses_api=True,
)
messages = [
SystemMessage(
content="Translate the following from English into Italian"
),
HumanMessage(content="hi!"),
]
model.invoke(messages).pretty_print()
```
```output
================================== Ai Message ==================================
Ciao!
```
To use `init_chat_model` you must set the `AZURE_AI_PROJECT_ENDPOINT`, and (optional) `OPENAI_API_KEY` environment variables. Use the provider `azure_ai`:
```python
from langchain.chat_models import init_chat_model
from dotenv import load_dotenv
load_dotenv()
# Option A) Using project endpoint
os.environ["AZURE_AI_PROJECT_ENDPOINT"] = "https://{your-resource-name}.services.ai.azure.com/api/projects/<project>"
# Option B) Using OpenAI endpoint
os.environ["OPENAI_API_BASE"] = "https://{your-resource-name}.services.ai.azure/openai/v1"
os.environ["OPENAI_API_KEY"] = "{your-key}"
model = init_chat_model("azure_ai:gpt-5-mini")
```
### Microsoft Foundry Agent Service
You can build multi agent graphs in LangGraph by using the integration with Microsoft Foundry Agent Service. The class `AgentServiceFactory` allows you to create agents and nodes that can be used to compose graphs.
```python
from azure.identity import DefaultAzureCredential
from langchain_core.messages import AIMessage, HumanMessage
from langchain_azure_ai.agents import AgentServiceFactory
from langchain_azure_ai.utils.agents import pretty_print
factory = AgentServiceFactory(
project_endpoint="https://{your-resource-name}.services.ai.azure.com/api/projects/{your-project}",
credential=DefaultAzureCredential()
)
echo_node = factory.get_agent_node(name="my-echo-agent", version="latest")
```
Agent Service nodes run in Microsoft Foundry but can be added to any graph:
```python
graph.add_node("expert_node", echo_node)
```
Use the graph as usual:
```python
agent = graph.compile()
messages = [HumanMessage(content="I'm a genius and I love programming!")]
response = agent.invoke({"messages": messages})
pretty_print(response)
```
```output
================================ Human Message =================================
I'm a genius and I love programming!
================================== Ai Message ==================================
Name: my-echo-agent
You're not a genius and you don't love programming!
```
# Contributing
We welcome contributions! Please see [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines on how to contribute to this project.