> ## Documentation Index
> Fetch the complete documentation index at: https://docs.meliai.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Use Meliai with LangChain for Sovereign AI Projects

> Connect LangChain to Meliai using the ChatOpenAI class with a custom base_url. All LangChain chains, agents, and memory features work without modification.

LangChain's `ChatOpenAI` class accepts a custom `base_url`, making Meliai a drop-in provider for any LangChain application. You keep every chain, agent, tool, and memory component exactly as-is — only the endpoint and API key change. Your data stays within the EU on every call.

<Steps>
  <Step title="Install LangChain">
    Install the `langchain-openai` package, which provides the `ChatOpenAI` and `OpenAIEmbeddings` classes used throughout this guide.

    ```bash theme={null}
    pip install langchain-openai
    ```
  </Step>

  <Step title="Configure ChatOpenAI with Meliai">
    Pass your Meliai API key and base URL when instantiating `ChatOpenAI`. Everything else — temperature, streaming, callbacks — works the same way.

    ```python theme={null}
    import os
    from langchain_openai import ChatOpenAI

    llm = ChatOpenAI(
        model="<MODEL_ID>",
        api_key=os.environ["MELIAI_API_KEY"],
        base_url="https://api.meliai.ai/v1",
    )

    result = llm.invoke("What are the benefits of European data sovereignty?")
    print(result.content)
    ```
  </Step>

  <Step title="Use in a chain">
    Compose your `ChatOpenAI` instance with prompts and other runnables using LangChain Expression Language (LCEL). No changes are needed compared to a standard OpenAI setup.

    ```python theme={null}
    from langchain_core.prompts import ChatPromptTemplate

    prompt = ChatPromptTemplate.from_messages([
        ("system", "You are a helpful assistant."),
        ("user", "{input}"),
    ])

    chain = prompt | llm
    result = chain.invoke({"input": "Name three open-weight LLMs."})
    print(result.content)
    ```
  </Step>

  <Step title="Use embeddings">
    `OpenAIEmbeddings` accepts the same `base_url` parameter, giving you Meliai-hosted embedding models for vector search, RAG pipelines, and semantic similarity tasks.

    ```python theme={null}
    from langchain_openai import OpenAIEmbeddings

    embeddings = OpenAIEmbeddings(
        model="<EMBEDDING_MODEL_ID>",
        api_key=os.environ["MELIAI_API_KEY"],
        base_url="https://api.meliai.ai/v1",
    )

    vector = embeddings.embed_query("European AI infrastructure")
    ```
  </Step>
</Steps>

<Tip>
  Append a routing suffix to any model ID to control how Meliai selects a provider. For example, use `<MODEL_ID>:eco` to prefer the lowest-carbon data center, or `<MODEL_ID>:speed` to prioritise the fastest available provider.
</Tip>