> ## 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.

# Migrate from OpenAI to Meliai – Two-Line Change Required

> Migrate from OpenAI to Meliai by changing base_url and your API key. All existing OpenAI SDK code continues to work without further changes.

Meliai implements the OpenAI API shape exactly — the same request fields, the same response structure, the same streaming format, and the same SDK interfaces. Migrating means changing two values: your API key and the `base_url`. Every other line of your existing code stays untouched.

<Steps>
  <Step title="Get a Meliai API key">
    Create a key at [meliai.ai/account/api/keys](https://meliai.ai/account/api/keys). It will be formatted as `sk-mel-<KEY>`.

    Store it in your environment alongside (or instead of) your existing OpenAI key:

    ```bash theme={null}
    export MELIAI_API_KEY="sk-mel-<YOUR_KEY>"
    ```

    <Tip>
      If some services in your stack still need the OpenAI key, keep both environment variables. You can reference `OPENAI_API_KEY` for those services and `MELIAI_API_KEY` for Meliai-routed calls.
    </Tip>
  </Step>

  <Step title="Update base_url and api_key">
    Open the file where you initialise your OpenAI client and make the following two-line change:

    ```python theme={null}
    import os
    from openai import OpenAI

    client = OpenAI(
        api_key="sk-openai-...",              # [!code --]
        api_key=os.environ["MELIAI_API_KEY"], # [!code ++]
        base_url="https://api.meliai.ai/v1", # [!code ++]
    )
    ```

    For Node.js projects the change is identical in shape:

    ```javascript theme={null}
    import OpenAI from "openai";

    const client = new OpenAI({
      apiKey: "sk-openai-...",                    // [!code --]
      apiKey: process.env.MELIAI_API_KEY,         // [!code ++]
      baseURL: "https://api.meliai.ai/v1",        // [!code ++]
    });
    ```

    That is the complete migration for SDK-based code. No other changes are required.
  </Step>

  <Step title="Pick a model">
    OpenAI model names (e.g. `gpt-4o`, `gpt-3.5-turbo`) don't map 1:1 to Meliai model IDs, because Meliai routes exclusively to open-weight models. Browse equivalent models by capability at [meliai.ai/hub](https://meliai.ai/hub), or list them programmatically:

    ```bash theme={null}
    curl https://api.meliai.ai/v1/models \
      -H "Authorization: Bearer $MELIAI_API_KEY"
    ```

    Substitute the model ID you choose into your `model` field:

    ```python theme={null}
    response = client.chat.completions.create(
        model="<MODEL_ID>",
        messages=[{"role": "user", "content": "Hello!"}],
    )
    ```
  </Step>

  <Step title="Optionally add a routing suffix">
    Meliai lets you influence how your request is routed across EU providers by appending a suffix to the model ID. No code changes beyond the suffix are needed.

    | Suffix      | Behaviour                                             |
    | ----------- | ----------------------------------------------------- |
    | `:balanced` | Default. Balances latency, cost, and availability.    |
    | `:speed`    | Routes to the fastest available provider.             |
    | `:price`    | Routes to the lowest-cost provider.                   |
    | `:eco`      | Routes to the highest renewable-energy provider.      |
    | `:batch`    | Queues the request for asynchronous batch processing. |

    ```python theme={null}
    response = client.chat.completions.create(
        model="<MODEL_ID>:eco",  # prefer renewable-energy providers
        messages=[{"role": "user", "content": "Summarise this document."}],
    )
    ```

    See the [Routing](/concepts/routing) page for the full details.
  </Step>
</Steps>

<Note>
  All OpenAI SDK features work identically through Meliai: streaming (`stream=True`), tool calling, vision (image inputs), structured output (`response_format`), and logprobs. You do not need to change any feature-specific code.
</Note>

## What you gain after migrating

Switching to Meliai gives you everything your existing OpenAI code already produces, plus:

* **EU data residency** — requests are routed only to providers inside the European Union.
* **GDPR compliance** — your data is never used for model training.
* **Environmental telemetry** — every response carries an `environment_impact` block with energy, carbon, water, and renewable-percentage figures for that specific call.
* **Auto-failover** — if one EU provider is unavailable, Meliai automatically retries on another, with no change to your code.